**An Exploration of the Roles of Ferric Iron Chelation-Strategy Components in the Leaves and Roots of Maize Plants**

**Georgios Saridis 1, Styliani N. Chorianopoulou 2,\*, Yannis E. Ventouris 2, Petros P. Sigalas <sup>3</sup> and Dimitris L. Bouranis <sup>2</sup>**


Received: 25 February 2019; Accepted: 16 May 2019; Published: 18 May 2019

**Abstract:** Plants have developed sophisticated mechanisms for acquiring iron from the soil. In the graminaceous species, a chelation strategy is in charge, in order to take up ferric iron from the rhizosphere. The ferric iron chelation-strategy components may also be present in the aerial plant parts. The aim of this work was to search for possible roles of those components in maize leaves. To this end, the expression patterns of ferric iron chelation-strategy components were monitored in the leaves and roots of mycorrhizal and non-mycorrhizal sulfur-deprived maize plants, both before and after sulfate supply. The two levels of sulfur supply were chosen due to the strong impact of this nutrient on iron homeostasis, whilst mycorrhizal symbiosis was chosen as a treatment that forces the plant to optimize its photosynthetic efficiency, in order to feed the fungus. The results, in combination with the findings of our previous works, suggest a role for the aforementioned components in ferric chelation and/or unloading from the xylem vessels to the aerial plant parts. It is proposed that the gene expression of the DMA exporter *ZmTOM1* can be used as an early indicator for the establishment of a mycorrhizal symbiotic relationship in maize.

**Keywords:** arbuscular mycorrhizal symbiosis; iron deficiency; iron homeostasis; sulfur deficiency; sulfur interactions; *Zea mays*

### **1. Introduction**

Iron is an essential element for plant growth and productivity. It participates in electron transfer reactions and is central to the function of heme- and Fe–S cluster-requiring enzymes; thus, iron is required for various cellular processes, including respiration, photosynthesis, sulfur assimilation, and nitrogen fixation [1]. Although soil contains abundant iron, its bioavailability is extremely low. Under aerobic conditions and in the physiological pH range, iron is mainly present as oxidized ferric (hydr)oxides, which are sparingly soluble and not available to plants [2]. Therefore, plants have developed sophisticated and tightly regulated mechanisms for acquiring iron from soil. Non-graminaceous plants reduce ferric iron to the more soluble ferrous form at the root surface, by exporting a number of metabolites (including organic acids, phenolics, flavonoids and flavins), inducing the expression of ferric-chelate reductase, and they transport the resulting ferrous ions across the root plasma membrane. In contrast, graminaceous plants use a chelation strategy to take up iron from their rhizosphere. They secrete phytosiderophore compounds from their roots, phytosiderophores solubilize iron in the soil,

and then the roots take up the resulting ferric iron–phytosiderophore complexes [1]. Maize in particular biosynthesizes only the 2- -deoxymugineic acid (DMA) via DMA synthase (ZmDMAS1). DMA is secreted into the rhizosphere via a specific exporter (ZmTOM1), and then the ferric iron-DMA complex is taken up by the ZmYS1 transporter [3].

Once iron has entered the root symplast, it is transported through the root cortex in the form of ferrous iron–nicotianamine complexes. After entering the root pericycle, it can be loaded into the xylem for transport to the shoot's sink tissues. The dominant form of iron in the xylem sap of non-graminaceous plants is ferric iron–citrate, whereas in graminaceous plants the bound iron forms may be ferric iron–citrate, along with ferric iron–phytosiderophores [4]. An important sink tissue for iron is the leaves, where it re-enters the symplast, is reduced to the ferrous form, and is again found as ferrous iron–nicotianamine.

During this "route of iron" within a graminaceous plant, it seems that ferric iron chelation-strategy components may also be present in the aerial plant parts. The expression of the DMA synthase gene in the leaves of rice and barley suggests a possible role(s) of DMA in leaf iron homeostasis, under various growth conditions and developmental stages. In rice shoot tissue, no expression of *OsDMAS1* was detected in the leaves of iron-sufficient plants, whereas under iron-deficient conditions, the *OsDMAS1* gene was specifically expressed in vascular bundles of the leaves [3].

Therefore, a plausible question arises as regards the role (or roles), if any, of those components in maize leaves. In a previous work of our group it was demonstrated that arbuscular mycorrhizal (AM) symbiosis impeded the expected iron deprivation responses in sulfur-deprived maize plants [5]. Sulfur deficiency has a negative impact on the iron homeostasis of all plants, due to the fact that the primary precursor of nicotianamine (i.e., the primary iron-chelating compound into the symplasm) is the sulfur-containing amino acid methionine. Nicotianamine is synthesized via the enzyme nicotianamine synthase (NAS), which uses S-adenosyl-methionine as a substrate molecule. For the graminaceous plants in particular, sulfur deficiency also exerts an impact on the iron uptake, since nicotianamine is the precursor compound of phytosiderophores. In this study, the gene expression patterns of ferric iron chelation-strategy components were monitored in the leaves (*ZmDMAS1*, *ZmTOM1* and *ZmYS1*) and roots (*ZmDMAS1* and *ZmTOM1*) of mycorrhizal and non-mycorrhizal sulfur-deprived maize plants, both before and after (24 and 48 h) sulfate supply.

### **2. Results**

### *2.1. Expression Levels of ZmDMAS1 in the Leaves*

No significant difference in the expression levels of *ZmDMAS1* in the leaves of non-mycorrhizal (NM) plants was monitored during the long sulfur-deficient period of time. The sulfur supply on day 60 resulted in a transient upregulation of this gene 24 h after this supply, whilst afterwards the expression levels reverted to the previous values (Figure 1a,c). The leaves of mycorrhizal (M) plants showed a differential response, both during the sulfur-deprived as well as after the sulfur-repletion period. The expression levels were constantly increased during the first 60 days of the study. The addition of sulfate had no significant effect on the expression values, which remained unaffected on days 61 and 62 (Figure 1b,d).

### *2.2. Expression Levels of ZmTOM1 in the Leaves*

A prolonged sulfur deprivation resulted in a strong overexpression of *ZmTOM1* in the leaves of NM plants on day 60. The influx of sulfate caused a further overexpression 24 h after the addition of sulfur. The second day after the sulfur supply, the expression levels decreased significantly (Figure 2a,c). *ZmTOM1* was also strongly upregulated in the leaves of M plants on day 60, but in this case, it was already upregulated from day 45. The sulfur supply induced a strong overexpression on day 61, followed by an equally strong downregulation on day 62 (Figure 2b,d).

**Figure 1.** Expression of *ZmDMAS1* in the leaves of non-mycorrhizal (NM; **a**,**c**) and mycorrhizal (M; **b**,**d**) maize plants, (**a**,**b**) before and (**c**,**d**) after the sulfur supply, relative to the expression of ubiquitin. Bars show the mean of the biological replicates ± SE, asterisk (\*) indicates the statistically significant difference between the sampling and the respective control at *p* < 0.05.

**Figure 2.** Expression of *ZmTOM1* in the leaves of non-mycorrhizal (NM; **a**,**c**) and mycorrhizal (M; **b**,**d**) maize plants, (**a**,**b**) before and (**c**,**d**) after the sulfur supply, relative to the expression of ubiquitin. Bars show the mean of the biological replicates ± SE, asterisk (\*) indicates the statistically significant difference between the sampling and the respective control at *p* < 0.05.

### *2.3. Expression Levels of ZmYS1 in the Leaves*

*ZmYS1* was strongly upregulated in the leaves of NM plants on day 60, before sulfur supply. The sulfur supply resulted in a transient downregulation on day 61, followed by a strong overexpression 2 days after the influx of sulfate (Figure 3a,c). The expression levels of this gene in the leaves of M plants increased constantly during the sulfur-deprived period and presented a strong overexpression on both sampling days. The addition of sulfate resulted in a strong downregulation on day 61, whilst the expression levels also remained reduced on day 62 (Figure 3b,d).

**Figure 3.** Expression of *ZmYS1* in the leaves of non-mycorrhizal (NM; **a**,**c**) and mycorrhizal (M; **b**,**d**) maize plants, (**a**,**b**) before and (**c**,**d**) after the sulfur supply, relative to the expression of ubiquitin. Bars show the mean of the biological replicates ± SE, asterisk (\*) indicates the statistically significant difference between the sampling and the respective control at *p* < 0.05.

### *2.4. Expression Levels of ZmDMAS1 in the Roots*

No significant difference in the expression levels of *ZmDMAS1* was monitored in the roots of NM plants during the sulfur deprivation treatment. The sulfur supply induced a strong downregulation of the expression, followed by an overexpression 2 days after the sulfate influx (Figure 4a,c). No significant difference in the expression levels of *ZmDMAS1* during sulfur deprivation was also observed in the roots of M plants. The addition of sulfate caused a strong downregulation of this gene expression on day 61. The same reduced levels were also observed on day 62 (Figure 4b,d).

### *2.5. Expression Levels of ZmTOM1 in the Roots*

The expression levels of *ZmTOM1* in the roots of NM plants were significantly downregulated on day 45 and remained reduced until day 60. The sulfur supply induced a further downregulation on day 61, whilst the next day the expression was strongly upregulated (Figure 5a,c). In the roots of M plants, the respective expression levels of *ZmTOM1* were constantly downregulated, both before as well as after the sulfur supply (Figure 5b,d).

**Figure 4.** Expression of *ZmDMAS1* in the roots of non-mycorrhizal (**NM**; **a**,**c**) and mycorrhizal (**M**; **b**,**d**) maize plants, (**a**,**b**) before and (**c**,**d**) after the sulfur supply, relative to the expression of ubiquitin. Bars show the mean of the biological replicates ± SE, asterisk (\*) indicates the statistically significant difference between the sampling and the respective control at *p* < 0.05.

**Figure 5.** Expression of *ZmTOM1* in the roots of non-mycorrhizal (NM; **a**,**c**) and mycorrhizal (M; **b**,**d**) maize plants, (**a**,**b**) before and (**c**,**d**) after the sulfur supply, relative to the expression of ubiquitin. Bars show the mean of the biological replicates ± SE, asterisk (\*) indicates the statistically significant difference between the sampling and the respective control at *p* < 0.05.

### *2.6. Data Meta-Analysis*

The data on all the iron homeostasis components examined in this study, as well as in our previous works [5,6] were combined and further analyzed, and the comparative analysis is depicted in Table 1. In this analysis, the relative expression ratios of *ZmNAS3*, *ZmNAS1*, *ZmDMAS1*, *ZmTOM1* and *ZmYS1* in the leaves and roots of M plants were calculated using the respective values of NM plants as control. Even on day 30, the ferric iron chelation components were differentially regulated in the leaves of

M plants, although there was no difference observed in the M roots on that day, except for a strong downregulation of *ZmTOM1*. As a matter of fact, *ZmTOM1* was the sole iron homeostasis component permanently downregulated in the roots and upregulated in the leaves of M plants throughout the experiment. On the other hand, all of the examined genes were downregulated in the M roots after the addition of sulfate to the nutrient solution.

### **3. Discussion**

### *3.1. Roles of Ferric Iron Chelation Components in Maize Leaves*

Following its entry in the root central cylinder, iron is transported via the xylem vessels to be utilized in the aerial parts. The dominant bound iron forms in the xylem sap of graminaceous plants were found to be of two types: ferric iron-citrate, along with various ferric iron-phytosiderophores; DMA and citrate were present in large concentrations in the xylem sap from rice and maize [4]. The iron deficiency-inducible genes of barley have been found to be expressed almost exclusively in roots, whereas many iron deficiency-inducible genes in rice were expressed in both roots and shoots [3]. In the shoot tissues, the *OsDMAS1* promoter-GUS analysis showed an expression in vascular bundles, specifically under iron-deficient conditions. No GUS expression was detected in the leaves of iron-sufficient plants, whereas under iron-deficient conditions, a GUS activity was detected in the phloem sieve tubes and companion cells, as well as in the xylem parenchyma cells of the large vascular bundles [3]. A similar study assessing the tissue specific localization of *OsTOM1* showed the same GUS reporter gene expression pattern in the vascular bundles of the leaves of iron-deficient plants [7]. In the case of YS1, the expression of *HvYS1* was only specifically induced by iron deficiency in barley roots, whereas *ZmYS1* was expressed in maize in the leaf blades and sheaths, crown, and seminal roots [8,9]. The fact that ZmYS1 was expressed in the leaves led to the hypothesis that it must have additional functions that are not related to iron uptake from the soil. The expression of ZmYS1 was only found in the leaves of iron deficient maize plants, which are thus producing DMA. In those iron-starved plants, the ferric iron-DMA substrate for ZmYS1 could arrive to the leaves following transport from the roots through the xylem.

In this study, M maize plants presented an early response in regulating iron homeostasis in the young expanding leaves (Table 1). The young expanding leaves were chosen as strong sinks of iron, whilst AM symbiosis was chosen as a treatment that forces the plant to optimize its photosynthetic efficiency, in order to feed the fungus. Ferric iron chelation components were differentially regulated in the leaves of M plants even on day 30, when the AM symbiosis was not yet visibly established. The roots of the maize plants were colonized by the fungus on day 45. In view of this, on day 30 the plant was already prepared for the upcoming symbiosis, and its efforts to maintain an efficient iron homeostasis in the young leaves was evident throughout the experimental period (Table 1). However, it seems that the genes related to iron uptake from the roots do have a role in the aerial plant parts. More specifically, it is hypothesized that the enzymes ZmNAS1, ZmDMAS1, ZmTOM1 and ZmYS1 have a role in the ferric iron chelation and/or unloading from the xylem vessels.

It is hypothesized that DMA synthesized in the xylem parenchyma cells (through *ZmNAS1* and *ZmDMAS1*) is secreted to the xylem vessels (through *ZmTOM1*). The xylem sap in maize may contain ferric iron-citrate and ferric iron-DMA coming up from the roots. The secreted DMA chelates ferric iron delivered there in the form of ferric iron-citrate, because the ferric iron-DMA chelate has a higher stability than the ferric iron-citrate [10]. The resulting ferric iron-DMA, as well as the root-originated ferric iron-DMA in the xylem, may then enter the neighboring xylem parenchyma cells via the ZmYS1 transporter. The chelated iron may then be transported throughout the leaf and plant body via YSL transporters, for an efficient iron remobilization (Figure 6).

**Table 1.** Relative expression ratios of *ZmNAS3*, *ZmNAS1*, *ZmDMAS1*, *ZmTOM1* and *ZmYS1* in the leaves and roots of mycorrhizal plants, before (days 30, 45, and 60) and after (days 61 and 62) the sulfur supply, relative to the expression of ubiquitin. Non-mycorrhizal plants were used accordingly as a control for the calculation of the relative expression ratios. The values show the mean of the biological replicates ± SE. Table cells highlighted with dark grey indicate an upregulation, and those in light grey indicate a downregulation of the respective gene, when the difference between the sampling and the respective control was statistically significant at *p* < 0.05.


**Figure 6.** Conceptual model of iron xylem transport in young maize leaves, xylem-to-phloem iron exchange and iron re-translocation to younger tissues. The processes are considered to take place in the large vein of a maize leaf: (**a**) xylem vessel, (**b**) xylem parenchyma cell, and (**c**) phloem. DMA: deoxymugineic acid, DMAS: DMA synthase, NA: nicotianamine, NAS: nicotianamine synthase, SAM: S-adenosyl-methionine, TOM: DMA efflux transporter, YS: yellow stripe transporter, and YSL: yellow stripe like transporter.

### *3.2. Arbuscular Mycorrhizal Fungus Provides Iron to M Plants*

The expression patterns of the examined genes were completely different among the various treatments, i.e., NM vs. M under sulfur deprivation and NM vs. M after sulfate supply. This fact holds true for the roots as well as for the leaves of the examined maize plants. This can be explained if we consider the hypothesis that NM plants suffer from iron deficiency throughout the treatment, whilst mycorrhizal colonization prevented iron deprivation responses due to a symbiotic iron uptake pathway [5].

The expression profile patterns of the examined genes in the roots of M and NM maize plants, both before as well as after sulfur supply, supported the previously suggested hypothesis that the fungus provides iron to the plants [5]. Despite the facts that: (a) before the addition of sulfur the concentration of total Fe in the shoots of NM plants decreased from day 45 to day 60, whilst the corresponding concentrations in M plants remained at the same levels and (b) five days after the addition of sulfur the concentration of total iron in M plants was higher than the corresponding concentration in NM plants [5], none of the genes related to iron uptake was overexpressed in the roots of M plants, neither during the sulfur deprivation period, nor after the supply of sulfur. Given this, in the M plants iron was probably delivered to the central cortex through the fungal arbuscules. Consequently, the M plants did not need to excrete DMA to their rhizosphere and *ZmTOM1* was strongly downregulated (Table 1, Figures 4 and 5, [5]).

Twenty four hours after the sulfur supply, a general transient downregulation of all the genes related to iron uptake was monitored in the roots of both M and NM plants (Figures 4 and 5, [5,6]). Sulfate may act as a signal molecule regulating the expression of the iron uptake related genes in roots. It is suggested that the "massive entry" of inorganic sulfur into the root cells creates a transient "shortage of organic sulfur" condition, which is required for the DMA synthesis and that during this period of time the plants downregulate the whole DMA biosynthesis pathway, until organic sulfur will be again available.

Two days after the addition of sulfur to the nutrient solution, a diverse response was observed: the iron-uptake-related genes were upregulated in the roots of NM plants, whilst their expression in the M roots remained stable and downregulated from day 61 (Table 1, Figures 4 and 5, [5]).

Moreover, *ZmTOM1* was the only gene permanently overexpressed in the leaves and downregulated in the roots of M plants relative to NM ones, throughout the long-term experiment, even on day 30 when the M plants were not actually colonized by the fungus yet (Table 1). Therefore, the expression of *ZmTOM1* could be an early indicator of a mycorrhizal symbiotic relationship.

### **4. Materials and Methods**

### *4.1. Plant Material and Growth Conditions*

Mycorrhizal (M) and non-mycorrhizal (NM) maize plants were grown in pots with sterile river sand and practically insoluble FePO4 (500 mg per pot) in a long-term experiment, as previously described [5]. The plants were watered with a nutrient solution deprived of iron and sulfur and containing a minimum phosphorus concentration, in order to enhance the establishment of the symbiotic relationship with the AM fungus *Rhizophagus irregularis*. Iron was provided to plants throughout the experiment in the sparingly soluble form of FePO4. After a 60-day period of sulfur deprivation, sulfur was provided to the plants in the form of sulfate (2.5 mM CaSO4 2H2O and 1 mM MgSO4 7H2O).

### *4.2. Plant Samplings*

The samplings were performed on days 30, 45, 60, 61 (24 h after the sulfur supply) and 62 (48 h after the sulfur supply) after sowing and 3 h after the onset of light. The sampling of day 60 took place before the addition of sulfur. The lateral roots, as well as two young expanding leaves, were immediately frozen in liquid nitrogen and stored at −80 ◦C until use. In each experiment, plant material from at least three biological replicates per treatment and sampling day was used [5].

### *4.3. Gene Expression Analysis*

The gene expression analysis was conducted by means of Real-Time RT-PCR, as previously described [5]. The oligonucleotide primers used for RT-qPCR were as previously referred to [6]. The efficiency of each Real-Time RT-PCR reaction was calculated using the LinRegPCR software [11]. The following mathematical formula, from [12], was used for the calculations of the relative expression

ratios of the target genes (*ZmDMAS1*, *ZmTOM1* and *ZmYS1*), whilst ubiquitin (*ZmUBQ*) was used as the reference gene:

$$ratio = \frac{\text{Exterge}^{\Delta C \text{Target}(control - sample)}}{\text{Err}^{\Delta C \text{Pref}(control - sample)}} \tag{1}$$

The samples of day 30 (for the samplings before the sulfur supply) or day 60 (for the samplings after the sulfur supply) of the respective treatment (NM or M) were used as control, unless otherwise specified.

### *4.4. Data Meta-Analysis*

Data from [5,6] were combined and further analyzed in order to make a comparative analysis, taking into consideration the data of all of the examined iron homeostasis components. The results of this analysis are depicted in Table 1.

### *4.5. Statistical Analysis*

The experiment was performed two times under the same conditions and during two distinct time periods: autumn 2013 and spring 2014. The data were analyzed by t-test variance analysis with a two-tailed distribution and two-sample unequal variance to determine the significance of differences among the samplings.

### **5. Conclusions**

It is suggested that the components related to iron uptake from the roots have a role in the ferric iron chelation and/or unloading from the xylem vessels in the aerial plant parts. Moreover, the expression of the gene codes for the DMA exporter *ZmTOM1* could be an early indicator for the establishment of a mycorrhizal symbiotic relationship.

**Author Contributions:** G.S., S.N.C. and D.L.B. conceived and designed the experiment, elaborated the research questions, analyzed the data, and wrote the article; G.S., P.P.S. and Y.E.V. carried out the gene expression analysis. All authors have read and approved the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Contribution of Root Hair Development to Sulfate Uptake in** *Arabidopsis*

### **Yuki Kimura 1, Tsukasa Ushiwatari 1, Akiko Suyama 1, Rumi Tominaga-Wada 2, Takuji Wada <sup>2</sup> and Akiko Maruyama-Nakashita 1,\***


Received: 26 February 2019; Accepted: 17 April 2019; Published: 19 April 2019

**Abstract:** Root hairs often contribute to nutrient uptake from environments, but the contribution varies among nutrients. In *Arabidopsis*, two high-affinity sulfate transporters, SULTR1;1 and SULTR1;2, are responsible for sulfate uptake by roots. Their increased expression under sulfur deficiency (−S) stimulates sulfate uptake. Inspired by the higher and lower expression, respectively, of *SULTR1;1* in mutants with more (*werwolf* [*wer*]) and fewer (*caprice* [*cpc*]) root hairs, we examined the contribution of root hairs to sulfate uptake. Sulfate uptake rates were similar among plant lines under both sulfur sufficiency (+S) and −S. Under −S, the expression of *SULTR1;1* and *SULTR1;2* was negatively correlated with the number of root hairs. These results suggest that both −S-induced *SULTR* expression and sulfate uptake rates were independent of the number of root hairs. In addition, we observed (1) a negative correlation between primary root lengths and number of root hairs and (2) a greater number of root hairs under −S than under +S. These observations suggested that under both +S and −S, sulfate uptake was influenced by the root biomass rather than the number of root hairs.

**Keywords:** *Arabidopsis thaliana*; root hair; sulfate transporter; sulfate uptake; CPC; CAPRICE; WER; WERWOLF

### **1. Introduction**

Plant nutrients are generally absorbed from the roots in appropriate chemical forms. From the root surface to cortex, nutrients are transported via the symplasmic or apoplastic routes, but have to take the symplasmic route when they have reached the endodermis. At the endodermis, nutrients are subsequently transported to the xylem parenchyma cells and loaded into the xylem stream for further transport to shoot tissues [1,2]. The root surface cell layers, epidermis, and cortex all play important roles in nutrient uptake by expressing high-affinity transporters for most essential nutrients [2–5]. It has been suggested that increasing the root surface area exposed to the soil environments enables root hairs to also contribute to nutrient uptake [6–8]. Root hairs have been shown to increase the uptake of some nutrients, including phosphate and ammonium [8–10], but not others, such as iron and silicon [11,12]. Root hair growth is stimulated by the deficiency of phosphorous, manganese, iron, and some other nutrients in *Arabidopsis*, whereas deficiencies of other nutrients fail to produce a similar phenotype [3,9,13–17]. Thus, the importance of root hairs in nutrient uptake varies among nutrients.

Plants absorb sulfate as the main source of sulfur through the activity of sulfate transporters (SULTRs) [5,18]. Sulfate taken up from the soil is transported to the shoots and primarily metabolized through reductive assimilation to form cysteine (Cys). Then, methionine and glutathione (GSH) are synthesized from Cys [5,18–20]. In *Arabidopsis*, two high-affinity sulfate transporters, SULTR1;1 and SULTR1;2, which are localized to the root epidermis and cortex, are responsible for the initial uptake of sulfate from the soil environment [21–25]. The transcript levels of both *SULTR1;1* and *SULTR1;2* are increased by sulfur deficiency (−S), which increases the sulfate uptake rate [21–25]. The −S-induced the expression of *SULTR1;1* and *SULTR1;2* is controlled by the promoter activities of their 5- -upstream regions [26–29].

Although SULTR1;1 and SULTR1;2 both contribute to sulfate uptake from the roots and their expression is stimulated by −S, they differ with regard to several features related to transcriptional regulation. When sulfate is sufficiently supplied, the transcript levels of *SULTR1;2* are higher than those of *SULTR1;1* are [24,25]. Under–S conditions, the transcript levels of *SULTR1;2* are still higher than those of *SULTR1;1*, but the rate of induction by −S is less in *SULTR1;2* than that in *SULTR1;1* [23–25,30,31]. The sulfate uptake rate and the levels of Cys and GSH in *SULTR1;1* and *SULTR1;2* knockout lines indicate that SULTR1;2 is the main contributor to sulfate uptake under sulfur sufficiency (+S) and −S conditions [23–25,30]. In addition, the tissue specificity of *SULTR1;1* and *SULTR1;2* expression in root surface cell layers is not identical: *SULTR1;1* is mainly expressed in the epidermis, including root hairs, whereas *SULTR1;2* is mainly expressed in the cortex [21,24,26].

The genetic control of root hair development has been extensively studied in *Arabidopsis* [6,7,9,32–35]. Root epidermal cells are separated into root hair and non-hair cells in *Arabidopsis*. Many transcription factors function in the early events of epidermal cell differentiation. In non-hair cells, a core transcriptional complex consisting of TRANSPARENT TESTA GLABRA1 (TTG1, WD40 repeat protein, [36]), GLABRA3 (GL3, bHLH transcription factor, [37]), ENHANCER of GLABRA3 (EGL3, bHLH transcription factor, [38]), and WERWOLF (WER, R2R3 MYB transcription factor, [39]) promotes the expression of the transcription factor GLABRA2 (GL2, homeobox-leucine zipper protein, [40]) which negatively regulates root hair cell-specific genes and positively regulates non-hair cell specific genes to determine non-hair cell fate. In root hair cells, a different protein complex, comprising R3 MYB transcription factors CAPRICE (CPC, [41]), TRIPTYCHON (TRY, [42]), and ENHANCER of TRY and CPC1 (ETC1, [43,44]), inhibits the association of the WER protein with the transcriptional complex described above, thereby repressing GL2 [45,46]. Therefore, given their functions in root hair development, mutation in WER results in an increase in the number of root hairs and mutation in CPC results in a decrease in the number of root hairs.

In addition to these transcription factors, which function as cell fate determinants, many other factors involved in root hair growth have been identified [35]. Among them, a basic Helix-Loop-Helix (bHLH) transcription factor, ROOT HAIR DEFECTIVE 6 (RHD6), plays key roles in determining root hair identity under the control of CPC and GL2 [35,47–49]. RHD6 promotes the expression of other bHLH transcription factors including RHD6-LIKE4 (RSL4), which positively regulates root hair growth [50]. The expression of RSL4 is also induced by ethylene through direct induction by the transcription factors ETHYLENE INSENSITIVE 3 (EIN3) and its homolog EIN3-LIKE1 (EIL1) [17]. EIN3 and EIL1 physically interact with RHD6 to synergistically induce *RSL4* expression [17]. As phosphate starvation increases the level of EIN3 protein [16], a phosphate starvation-responsive increase in the number of root hairs would also be mediated by RSL4 and the MYB transcription factor PHOSPHATE STARVATION RESPONSE 1 (PHR1) which plays a central role in phosphate starvation responses [51,52].

In the present study, we investigated the role of root hairs in sulfate uptake using *werwolf* (*wer*) and *caprice* (*cpc*) *Arabidopsis* mutants, as CPC and WER control epidermal cell fate upstream of the signal transduction system of root hair development [6,7,9,32–35]. Our results indicated that the fate of root epidermal cells influenced the expression of *SULTR1;1* and *SULTR1;2*. However, the –S-induced expression of *SULTR1;1* and *SULTR1;2,* and sulfate uptake were independent of root epidermal cell development. Overall, the sulfate uptake rate was likely influenced by the total root biomass rather than the number of root hairs under both +S and –S conditions.

### **2. Results**

### *2.1. SULTR1;1 and SULTR1;2 Were Di*ff*erentially Expressed in the Root Epidermal Cells under* −*S*

The difference in tissue-specific expression between *SULTR1;1* and *SULTR1;2* in the root surface was investigated by assessing green fluorescent protein (GFP) fluorescence in the root elongation zones of *PSULTR1;1-GFP* [26] and *PSULTR1;2-GFP* [27] plants. When the plants were grown under +S sulfur sufficient (S1500) conditions, GFP fluorescence was not detected in either plant line (data not shown). Under −S sulfur deficient (S15) conditions, GFP fluorescence was clearly visible in root elongation zones (Figure 1). In *PSULTR1;1-GFP* plants, GFP fluorescence was mainly detected in root hair cells (Figure 1a–c), whereas it was detected in non-hair cells in *PSULTR1;2-GFP* plants (Figure 1g–i). Interestingly, GFP fluorescence was detected in a striped pattern in the root hair cells and non-hair cells in *PSULTR1;1-GFP* and *PSULTR1;2-GFP* plants, respectively, suggesting that they were differentially expressed in these cell lines and played different roles in sulfate uptake from the root surface.

**Figure 1.** Expression of green fluorescent protein (GFP) in the root elongation zones of *PSULTR1;1-GFP* plants (**a**–**f**, [26]) and *PSULTR1;2-GFP* plants (**g**–**l**, [27]). Three *T2* generation lines carrying each construct were grown for 7 d on agar media supplied with 15 μM sulfate and observed using fluorescent microscopy (EVOS FL Auto 2). GFP images (**a**–**c**,**g**–**i**) and bright field images (**d**–**f**,**j**–**l**) are presented. Scale bars, 100 μm.

### *2.2. Primary Root Length and Numbers of Root Hairs Compensated for Each Other in wer and cpc*

To assess the relationship between sulfate uptake and root hair development, root phenotypes of *wer*, *cpc* and wild-type Columbia plants (WT) under S1500 and S15 conditions were observed (Figure 2). The number of root hairs 5 mm from the root tip were counted in plants grown under S1500 and S15 conditions (Figure 2a–c). The number of root hairs was significantly higher under S15 conditions than under S1500 conditions (Figure 2c), suggesting that plants increased the number of root hairs for more efficient sulfate uptake under –S. Under both sulfur conditions, *wer* had more root hairs than the WT, and *cpc* tended to have less root hairs than the WT, as reported previously [39,41] (Figure 2c).

**Figure 2.** Plant growth and root development of wild-type (WT), and *werwolf* (*wer*), and *caprice* (*cpc*) mutants grown under sulfur-sufficient and -deficient conditions. Plants were vertically grown for 10 d on agar media supplied with 1500 μM sulfate (sulfur sufficient, S1500, white bar) or 15 μM sulfate (sulfur-deficient, S15, black bar). (**a**) Images of plants on their media. (**b**) Typical root image of plants grown under the S15 condition. Scale bar, 500 μm. (**c**) Number of root hairs in root elongation zone. Number of root hairs 5 mm from root tip was counted. (**d**) Primary root lengths. (**e**) Shoot and root fresh weights (FW) per plant. Values and error bars indicate means ± SEM [n = 5 in (**c**), n = 8 to 12 in (**d**), and n = 3 in (**e**)]. Two-way ANOVA was used to detect the effects of sulfur conditions, genotypes, and their interactions. Asterisks indicate significant differences between the S1500 and S15 conditions (two-way analysis of vatiance [ANOVA], \* *P* < 0.05). As all comparisons in (**c**–**e**) did not detect interaction, the Tukey–Kramer test was used to analyze significant differences among mutants and the WT under the same growth condition. Different letters indicate significant differences (*P* < 0.05).

Primary root lengths were not influenced by sulfate concentrations (Figure 2d). Interestingly, primary root lengths were influenced in the opposite direction as the number of root hairs in the two mutants as compared to the WT. That is, primary root lengths tended to be shorter in *wer* and

longer in *cpc* than they were in WT under both the sulfur conditions. Significant negative correlations between the number of root hairs and primary root lengths were detected using Sperman's Rank correlation coefficient (R = 0.047, correlation constant = −0.89). These phenotypes suggested that plants adjusted their root biomass by balancing the inverse relationship between primary root lengths and root hair densities.

Both shoot and root fresh weights were decreased under –S conditions (Figure 2e). In *wer*, shoot fresh weights under S1500 and S15 conditions tended to be less than those in the WT (Figure 2e). When plants were grown under the S15 condition, the root fresh weights of *wer* and *cpc* tended to be less than those in the WT (Figure 2e).

### *2.3. Regulatory Gene Expression in Root Hair Development Was Consistent with Root Hair Numbers under* +*S but Not –S Conditions*

The number of root hairs was increased under –S conditions and, therefore, we analyzed the expression of transcription factors *CPC, WER, GL2, RHD6*, and *RSL4*, which regulate root hair development (Figure 3).

**Figure 3.** Expression of transcription factors regulating root hair development in wild-type (WT), *werwolf* (*wer)*, and *caprice* (*cpc*) mutants grown under S1500 and S15 conditions. (**a**) Relative transcript levels of *CPC, WER,* and *GL2,* and (**b**) those of *RHD6*, and *RSL4,* in root tissues were determined using quantitative reverse transcription-polymerase chain reaction (qRT-PCR). Plants were vertically grown for 10 d on S1500 (white bars) and S15 (black bars) agar media. Values and error bars indicate means ± SEM (n = 3). Two-way analysis of variance (ANOVA) was used to detect the effects of sulfur conditions, genotypes, and their interactions. In (**a**), all comparisons did not detect interactions, so the Tukey–Kramer test was applied to the three genotypes under the same growth conditions. In (**b**), interactions between sulfur conditions and genotypes were detected, so the Tukey–Kramer test was applied to all six experimental conditions. Different letters indicate significant differences (*P* < 0.05).

The expression of *CPC* in *wer* was higher under the S1500 condition but lower under the S15 condition than that in the WT, whereas the expression of *WER* was not influenced in *cpc* (Figure 3a). The expression of *GL2* was higher in *cpc* under both conditions than that in the WT.

The transcript levels of root hair-specific transcription factors, *RHD6* and *RSL4*, were higher in *wer* under the S1500 condition than they were in WT but were similar among the genotypes under S15 (Figure 3b). Their transcript levels tended to be lower in *cpc* under both conditions than they were in the WT, which was likely due to the increased expression of *GL2* in *cpc* as GL2 suppressed the expression of *RHD6* [49]. With regard to the expression of *RHD6* and *RSL4,* an interaction was found between sulfur conditions and genotypes using a two-way ANOVA, suggesting their expressions were differentially influenced by the disruption of WER or the number of root hairs under +S and −S.

### *2.4. Sulfate Uptake Was Constant among WT, wer, and cpc*

We analyzed the transcript levels of *SULTR1;1* and *SULTR1;2* in the roots and the sulfate uptake activity in WT, *wer,* and *cpc* plants grown under S1500 and S15 conditions (Figure 4). Sulfur deficiency increased the transcript levels of *SULTR1;1* and *SULTR1;2* in all plant lines (Figure 4a). Under the S1500 condition, transcript levels of *SULTR1;2* were similar between the mutants and WT, whereas those of *SULTR1;1* were higher in *wer* and lower in *cpc* than they were in WT (Figure 3a), which is consistent with previous reports [53,54]. These results indicated that the expression of *SULTR1;1* was correlated with the number of root hairs when there was sufficient sulfate for plant growth (Figure 2b,c and Figure 4a).

**Figure 4.** Transcript levels of *SULTR1;1* and *SULTR1;2,* and sulfate uptake activity in wild-type (WT), *werwolf* (*wer*), and *caprice* (*cpc*) mutants grown under S1500 and S15 conditions. (**a**) Relative transcript levels of *SULTR1;1* and *SULTR1;2* in root tissues determined using quantitative reverse transcription-polymerase chain reaction (qRT-PCR). (**b**) Sulfate uptake activity of plants. Absolute values of [35S] sulfate uptake rates are presented. Plants were vertically grown for 10 d on S1500 (white bars) and S15 (black bars) agar media. Values and error bars indicate means ± SEM [n = 3 in (**a**), n = 6 in (**b**)]. Two-way analysis of variance (ANOVA) was used to detect the effects of sulfur conditions, genotypes, and their interactions. Asterisks indicate significant differences between S1500 and S15 conditions (two-way ANOVA; \* *P* < 0.05). In (**a**), interactions between sulfur conditions and genotypes were detected and, so the Tukey–Kramer test was applied to all six experimental conditions. In (**b**), no interaction was detected and, so the Tukey–Kramer test was applied to the three genotypes grown under the same condition. Different letters indicate significant differences (*P* < 0.05).

However, under –S, transcript levels of *SULTR1;1* were lower in *wer* than they were in *cpc,* indicating a negative correlation with the number of root hairs (Figure 2b,c and Figure 4a). Supporting this observation, interactive effects between sulfur conditions and genotypes on the expression of *SULTR1;1* and *SULTR1;2* were detected using a two-way ANOVA, which indicates that the effects of the number of root hairs on their expression would be switched from positive to negative with environmental changes from +S to −S.

The sulfate uptake rate was stimulated under the S15 condition, which was consistent with previous reports [23–25,27,55,56] (Figure 4b). Despite the differences in *SULTR1;1* expression between the plant lines, sulfate uptake rates were relatively constant between the two lines under both S1500 and S15 conditions (Figure 4b).

The levels of sulfur-containing metabolites, sulfate, Cys, GSH and total sulfur in the precipitate after extraction with 10 mM HCl were decreased under the S15 condition in all plant lines (Figure 5). Similar to the sulfate uptake rate, the content of sulfate, Cys, and GSH in shoots was generally not influenced by mutations in *WER* and *CPC* under both S1500 and S15 conditions, except for slightly higher sulfur levels in the precipitate of *wer* under the S15 condition than in the WT.

**Figure 5.** Accumulation of sulfate, cysteine (Cys), glutathione (GSH), and sulfur in shoots of wild-type (WT), *werwolf* (*wer*), and *caprice* (*cpc*) mutants grown under S1500 and S15 conditions. Plants were vertically grown for 10 d on S1500 (white bars) and S15 (black bars) agar media. Sulfate content in shoot tissues was determined using ion chromatography. Cys and GSH content of shoot tissues was analyzed using a high-performance liquid chromatography (HPLC)-fluorescent detection system, after labeling thiol bases with monobromobimane. Total sulfur content in precipitates was analyzed using inductively coupled plasma-mass spectroscopy (ICP-MS) after nitric acid digestion. Values and error bars indicate means ± SEM (n = 3). Two-way analysis of variance (ANOVA) was used to detect effects of sulfur conditions, genotypes, and their interactions. Asterisks indicate significant differences between S1500 and S15 conditions (two-way ANOVA; \* *P* < 0.05). As all comparisons did not detect interactions, the Tukey–Kramer test was applied to detect significant differences among genotypes grown under similar conditions. Different letters indicate significant differences (*P* < 0.05).

The sum of sulfur levels analyzed in this study was converted to sulfur content in shoot per plant (Figure 6). Concentrations of sulfate, Cys, and GSH (Figure 5) were converted to sulfur content, and added to the sulfur content of the precipitate, and then the total was multiplied by the shoot FW per plant (Figure 2e). The sum of sulfur levels was lower in *wer* under the S1500 condition than it was in the WT and was similar among the genotypes under the S15 condition (Figure 6). Interaction between sulfur conditions and genotypes was detected using a two-way ANOVA, suggesting that sulfur levels were differentially controlled by the number of root hairs under +S and −S conditions.

**Figure 6.** Sum of sulfur levels in shoots of WT, *wer*, and *cpc* mutants grown under S1500 and S15 conditions. Plants were vertically grown for 10 d on S1500 (white bars) and S15 (black bars) agar media. Sulfate, Cys, GSH, and sulfur contents in precipitate of shoot tissues were calculated to μg sulfur per plant. The values and error bars indicate means ± SEM (n = 3). As two-way ANOVA detected the interaction between sulfur conditions and genotypes, the Tukey–Kramer test was applied to all six experimental conditions. Different letters mean significant differences (*P* < 0.05). Asterisks indicate significant differences between S1500 and S15 conditions (2-way ANOVA; \* *P* < 0.05).

### **3. Discussion**

The mechanisms by which cellular differentiation influences the cellular biochemical or physiological functions and, vice versa, how cellular functioninfluences cell fate, have bothlong been topics of discussion. In the case of sulfate uptake, two high-affinity sulfate transporters, SULTR1;1 and SULTR1;2, show different spatial expression patterns in root epidermal cells [21,24,26] (Figure 1). Specifically, we found that the expression of *SULTR1;1* was induced preferentially in the root-hair cells in a −S condition [57], whereas *SULTR1;2* was preferentially expressed in non-hair cells (Figure 1). Additionally, the expression of *SULTR1;1* was reported to be higher in *wer* and lower in *cpc* than in WT, whereas the expression of *SULTR1;2* was not influenced in the mutants [53,54]. Based on this information, we postulated that a decrease in the number of root hairs would reduce sulfate uptake activity. However, an increase or decrease in the number of root hairs did not influence the sulfate uptake rate in plants under either +S or −S conditions (Figure 4b). These results indicated that the number of root hairs does not contribute to the sulfate uptake rate, at least in the comparison among WT, *wer* and *cpc*.

Although we observed no difference in sulfate uptake between the WT and mutants, the contribution of root hairs to sulfate uptake cannot be completely ruled out because there was a negative correlation between the number of root hairs and primary root length [58] (Figure 2). The negative correlation suggested that there might be a mechanism to maintain a constant root biomass. A similar negative correlation was observed when plants were exposed to phosphate-deficient conditions, ethylene, or both [16,59], suggesting the involvement of general homeostatic mechanisms. The molecular machinery coordinating the negative interaction between root hair development and primary root length would be an interesting subject to investigate as a potential tool to improve the understanding of nutrient acquisition by plants. The S15 condition also induced a slight but significant increase in the number of root hairs (Figure 2c), which could contribute to the increase in sulfate uptake

rate under –S conditions. These results suggested that the sulfate uptake rate was influenced by the total root biomass rather than the number of root hairs under both +S and –S conditions.

The expression of both *SULTRs* was correlated with the number of root hairs negatively under the S15 condition, but positively under the S1500 condition (Figures 2c and 4a). A similar tendency was observed for the expression of *RHD6* and *RSL4* (Figure 3b). Under the S1500 condition, the expression levels of *RHD6* and *RSL4* were higher in *wer* and tended to be lower in *cpc* than they were in the WT, which was consistent with the number of root hairs (Figures 2c and 3b). However, under the S15 condition, the expression of *RHD6* and *RSL4* was similar among the plant lines, whereas the number of root hairs varied in the mutants in a manner similar to that under S1500 condition (Figures 2c and 3b). Although it is not clear whether the expression of *RHD6*, *RSL4,* and *SULTRs* was regulated by the same mechanism under –S conditions, these results suggested that another mechanism likely increased root hair development under –S conditions.

Furthermore, phosphate deficiency induces *RSL4* expression, probably by increasing the stability of EIN3 and EIL1, which both bind to cognate binding sites in the promoter of *RSL4* [16,17]. In another root hair-less line, NR23, several events are induced by phosphate deficiency, including increased the expression of phosphate transporter genes and secretion of acid phosphatases and organic acids, compared to the WT [8]. This observation also supports the differing contribution of root hairs to the uptake of phosphate and sulfate, which varies with the chemical forms of nutrients in the soil [3,4].

Although there were negative or positive correlations between the expression of *SULTR*s and the number of root hairs under both +S and −S conditions, the sulfate uptake rate did not fluctuate among the plant lines (Figure 4). This finding suggests that the transcript levels of *SULTR1;1* and *SULTR1;2* were not the only determinants of sulfate uptake rate. Further studies of the relationship between root architecture and sulfate uptake may shed light on the as yet unknown determinants of this process.

### **4. Materials and Methods**

### *4.1. Plant Materials and Growth Conditions*

*Arabidopsis thaliana* plants, ecotype "Columbia" (Col-0), were used as the wild-type (WT) while *cpc* [41] (*cpc-1*) and *wer* [39] (*wer-1*) mutants were obtained from the Arabidopsis Biological Resource Center (ABRC). Plants were grown at 22 ◦C under continuous light (40 μmol m−<sup>1</sup> s−1) conditions on mineral nutrient media containing 1% sucrose [60,61]. For the preparation of the agar medium, agar was washed twice with 1 L de-ionized water and vacuum filtered to remove the sulfate. Sulfur sufficient (S1500) agar medium was supplemented with 1500 μM MgSO4. Sulfur-deficient (S15) agar medium was supplemented with 15 μM MgSO4 and Mg concentration was adjusted to 1500 μM by adding MgCl2. After the indicated shown in each figure, the shoot and root tissues were harvested separately, rinsed with distilled water, and subsequently subjected to various analyses.

### *4.2. Observation of GFP Fluorescence*

The tissue-specific expression of GFP in *PSULTR1;1-GFP* and *PSULTR1;2-GFP* transgenic plants was visualized in whole mounts of 7-day-old plants using a fluorescent microscope system (EVOS FL Auto 2 Imaging System) equipped with the EVOS Light Cube, GFP (Ex: 470/22, Em: 525/50) (Thermo Fisher Scientific, USA).

### *4.3. Observation of Root Development*

The primary root lengths of plants were analyzed using images captured with a STAGE2000-BG system (AMZ System Science, Japan). The number of root hairs in 5 mm from the root tip was analyzed from the image captured using a CCD camera (WRAYCAM G500, WRAYMER, Japan) connected to a stereoscopic microscope (SW-700TD, WRAYMER). The free software package, ImageJ [62,63] was used for the analysis.

### *4.4. Quantitative Real-Time RT-PCR Analysis*

RNA preparation and RT were performed as reported previously [27,54,55]. Real-time PCR was carried out using a SYBR Green Perfect real-time kit (Takara, Japan) and Thermal Cycler Dice real-time system (Takara) using the gene-specific primers for CPC, CPC-F (5- -GGATGTATAAACTCGTTGG CGACAG-3- ) and CPC-R (5- -GCCGTGTTTCATAAGCCAATATCTC-3- ) [64]; for WER, WER-F (5- -TGGTAATAGGTATAACTTCATTTGC-3- ) and WER-R (5- -TTTGATTCCGAGTTTCTTACTAAGG ATG-3- ); for GL2, GL2-F (5- -TCGGATCACTGAGACCACAA-3- ) and GL2-R (5- -GTGTATCCCGG AACCAGTGT-3- ) [64]; for RHD6, RHD6-real-F (5- -TGATTTGGTGACAATGCTTGA-3- ) and RHD6-real-R (5- -GGAGAGAATGGCATCAATGG-3- ) [49]; for RSL4, RSL4\_q-PCR f new (5- -AACCTT GTGCCAAACGGGAC-3- ) and RSL4\_q-PCR r new (5- -CCAGGCCGTTGTAAGCCAAT-3- ) [17]; and, for *SULTR1;2*, SULTR1;2-1854F (5- -GGATCCAGAGATGGCTACATGA-3- ) and SULTR1;2-1956R (5- -TCGATGT CCGTAACAGGTGAC-3- ) [27]; for *SULTR1;1*, SULTR1;1-625F (5- -GCCATCACAA TCGCTCTCCAA-3- ) and SULTR1;1-750R (5- -TTGCCAATTCCACCCATGC-3- ) [30]; and for ubiquitin, UBQ2-144F (5- -CCAAG ATCCAGGACAAAGAAGGA-3- ) and UBQ2-372R (5- -TGGAGACGAGC ATAACACTTGC-3- ) [30]. The relative mRNA levels were calculated using ubiquitin2 as an internal standard.

### *4.5. Sulfate Uptake Assay*

Plants were vertically grown for 10 days on S1500 and S15 agar media. The roots were submerged in S1500 medium containing 15 μM [35S] sodium sulfate (American Radiolabeled Chemicals, USA) and incubated for 1 h. Washing and measurement were carried out as described previously [25,27,54,55,65,66].

### *4.6. Measurement of Sulfate, Cysteine and Glutathione, and Total Sulfur Levels*

Plant tissues were frozen in liquid nitrogen and homogenized in 5 volumes of 10 mM HCl using a Tissue Lyser (Retsch, Germany). After homogenization, the cell debris was removed by centrifugation, and the supernatant was subsequently analyzed.

For sulfate measurement, the extracts were diluted 100-fold with extra pure water and analyzed using ion chromatography (IC-2001, TOSOH, Japan). Using serial 30-μL injections, the diluted extracts were separated at 40 ◦C using a TSK SuperIC-AZ column (TOSOH) at a flow rate of 0.8 mL min−<sup>1</sup> with an eluent containing 1.9 mM NaHCO3 and 3.2 mM Na2CO3. Anion mixture standard solution 1 (Wako Pure Chemicals, Japan) was used as a standard.

Cys and GSH contents were determined using monobromobimane (Invitrogen, USA) labeling of the thiols after reduction of the plant extracts with dithiothreitol (DTT). The labeled products were then separated using HPLC (JASCO, Japan) using the TSKgel ODS-120T column (150 × 4.6 mm, TOSOH) and detected using a fluorescence detector, FP-920 (JASCO), monitoring for fluorescence of thiol-bimane adducts at 482 nm under excitation at 390 nm. Cys and GSH (Nacalai Tesque, Japan) were used as standards.

The total sulfur content was determined using inductively coupled plasma-mass spectroscopy (ICP-MS; Agilent7700, Agilent Technologies, USA). The precipitates obtained from the extraction described above were digested in 200 μL HNO3 at 95 ◦C for 30 min and then 115 ◦C for 90 min. After cooling to room temperature, the digested samples were diluted to 1 mL with extra pure water, filtered using 0.45 μm filters (DISMIC-03CP, ADVANTEC, Japan). The filtered samples were diluted 10 times with a solution consisting of 0.1 M HNO3 and 10 μg L−<sup>1</sup> gallium (KANTO CHEMICAL, Japan) as an internal standard, before being subjected to ICP-MS. Quantification was performed using the standard curve obtained via serial dilutions of the sulfur standard solution (KANTO CHEMICAL).

### *4.7. Statistical Analysis*

Two-way ANOVA was used to detect the effects of sulfur conditions, genotypes, and their interactions (Figures 2–6). Significant differences between S1500 and S15 conditions detected using a two-way ANOVA are indicated with asterisks (*P* < 0.05). Furthermore, where an interaction was detected between sulfur conditions and genotypes, the Tukey–Kramer test was applied to all experimental conditions (Figures 3 and 4a). In addition, where interaction was not detected between the values, the Tukey–Kramer test was used to analyze all genotypes grown under the same growth condition. Significant differences detected by the Tukey–Kramer test were shown with different letters (*P* < 0.05). Statcel4 software (OMS Publishing Inc., Tokyo, Japan) was used for all statistical analysis with the Microsoft Excel program.

**Author Contributions:** A.M.N., R.T.W. and T.W. designed the research. Y.K., T.U., A.S. and A.M.N. performed the experiments and analyzed the data. A.M.N. wrote the manuscript.

**Funding:** This work was supported by JSPS KAKENHI Grant Number 24380040 and 17H03785 (for A.M.N.).

**Acknowledgments:** We thank Yuki Mori (Kyushu University, Japan) for instructing the total sulfur analysis with ICP-MS. We thank Yasuo Niwa (University of Shizuoka, Japan) for providing the sGFP(S65T) vector. We thank Chihiro Iwaya and Yukiko Okuo for the technical support.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **CLE-CLAVATA1 Signaling Pathway Modulates Lateral Root Development under Sulfur Deficiency**

### **Wei Dong, Yinghua Wang and Hideki Takahashi \***

Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, MI 48824, USA; wdong@msu.edu (W.D.); wangyi64@msu.edu (Y.W.)

**\*** Correspondence: htakaha@msu.edu

Received: 2 March 2019; Accepted: 17 April 2019; Published: 18 April 2019

**Abstract:** Plant root system architecture changes drastically in response to availability of macronutrients in the soil environment. Despite the importance of root sulfur (S) uptake in plant growth and reproduction, molecular mechanisms underlying root development in response to S availability have not been fully characterized. We report here on the signaling module composed of the CLAVATA3 (CLV3)/EMBRYO SURROUNDING REGION (CLE) peptide and CLAVATA1 (CLV1) leucine-rich repeat receptor kinase, which regulate lateral root (LR) development in *Arabidopsis thaliana* upon changes in S availability. The wild-type seedlings exposed to prolonged S deficiency showed a phenotype with low LR density, which was restored upon sulfate supply. In contrast, the *clv1* mutant showed a higher daily increase rate of LR density relative to the wild-type under prolonged S deficiency, which was diminished to the wild-type level upon sulfate supply, suggesting that CLV1 directs a signal to inhibit LR development under S-deficient conditions. *CLE2* and *CLE3* transcript levels decreased under S deficiency and through CLV1-mediated feedback regulations, suggesting the levels of CLE peptide signals are adjusted during the course of LR development. This study demonstrates a fine-tuned mechanism for LR development coordinately regulated by CLE-CLV1 signaling and in response to changes in S availability.

**Keywords:** *Arabidopsis thaliana*; CLE peptide; CLAVATA1; root system architecture; small signaling peptide; sulfate; sulfur

### **1. Introduction**

Plant roots optimize nutrient uptake capacity by altering the root system architecture (RSA) in the soil environment [1]. Changes in nutrient availabilities have a distinct effect on RSA depending on the nutrient types and the amount supplied or locally concentrated in the soil environment [2]. The macronutrient sulfur (S), in the form of sulfate, is a mobile resource found in deeper soil profiles [1]. It is proposed that a combination of a thick and deep primary root (PR) with few and long lateral roots (LRs) can improve the uptake of S [3]. To achieve the adjustment of RSA, individual root traits can be regulated independently in response to changes in nutrient availabilities and patterns of nutrient distributions in the soil environment [4]. Among all root traits, LRs are phenotypically evaluated by length, total numbers, and density, which is considered a major determinant of RSA [5]. Changes in sulfate availability have a variable effect on LR development. For example, several studies have demonstrated that S deficiency leads to reduction in LR length [2,6–9] and LR number or density [2,7,9]. However, active growth of LR appears to be another response to sulfate starvation as it is described with longer LR length [10,11] and higher LR number or density [11–13].

Studies on functional characterization of small signaling peptides (SSPs) reveal that several distinct groups of SSPs play important roles in plant root development [14–24]. Nitrogen (N)-responsive C-TERMINALLY ENCODED PEPTIDEs (CEPs) are a group of SSPs known to be functional as negative regulators of LR development under N-limited conditions [14–17], while they have also been shown to be involved in long-distance regulation of N uptake [18]. A few distinct members of the CLAVATA3 (CLV3)/EMBRYO SURROUNDING REGION (CLE) family are also well characterized as SSPs regulating LR development in *Arabidopsis thaliana* (*A. thaliana*) [19,24]. *CLE1*, *-3,* -*4,* and *-7* are characterized particularly in relation to N nutritional responses as they are expressed in roots under N deficient conditions [19]. The LR phenotypes depicted in our previous study therefore highlight the *CLE3* gene expression enhanced in roots as a potential mechanism suppressing LR development under low N supply or in response to systemic N demand signals [19]. *CLE2* and *CLE3* also demonstrate positive responses of gene expression after N resupply to N-starved seedlings, as their transcript levels are shown to dramatically increase in response to nitrate and ammonium, respectively [19,25]. *CLE1, -2, -3, -4,* and *-7* are predominantly expressed in pericycle cells in roots, while *CLE1* and *CLE5* promoters are also found to be active in epidermal cells of the primary root tip [19]. These partially overlapping expression patterns, environmental responses and feedback regulation suggest functional redundancy of these N-responsive *CLE* genes in LR development. In addition to mechanisms characterized in relation to the N status, *CLE* genes are also known to be transcriptionally modulated by changes in availability of other macronutrients including S, phosphorus (P) and potassium (K) [24,26], as well as perturbation of cellular status caused by phytohormones and environmental stimuli [24,27]. More specifically to responses to S in roots, *CLE12* and *CLE2* are known to be up- and down-regulated, respectively, by S deprivation [24]. Thus, *CLE2* seems to be controlled by the S-responsive pathways in addition to being up-regulated by resupply of N [19]. In contrast, the S-responsive regulation of *CLE3* gene expression has not been studied despite its roles in LR development documented in relation to responses to N nutrition. CLE2 peptide has been shown to physically bind to the CLE receptor CLAVATA1 (CLV1) [28], and CLE3 requires CLV1 to transmit signals to modulate LR development [19]. Based on these aspects of nutrient-responsive regulation of *CLE2* and *CLE3* gene expression and their specific relationship with CLV1, we focused on investigating the effect of S on the CLE-CLV1 signaling pathway and LR development.

Here, we report the CLE-CLV1 signaling pathway is associated with S-responsive mechanisms modulating LR development in *A. thaliana*. The results shown in this study suggest a link between the morphological responses of LRs and the CLE-CLV1 signaling pathway, as well as S-responsive regulations of *CLE2* and *CLE3* genes in *A. thaliana* seedlings exposed to prolonged S deficiency.

### **2. Results**

### *2.1. CLAVATA1 Controls Lateral Root Development under S Deficiency*

To investigate the effect of S supply on root development, the wild-type *A. thaliana* (accession Columbia-0 (Col-0)) were germinated and precultured on a –S (15 μM sulfate) or +S (1500 μM sulfate) medium for 7 days. The seedlings were then transferred to the medium with the same concentration of sulfate, or from the –S preculture to the +S medium, or from the +S preculture to the –S medium, and grown for 3 days to validate the effect of S starvation and S replenishment (Figure 1). The most significant changes in root morphology at Day 10 were the decrease in length and number of LRs after long-term limitation of S (Figure 1a; plants transferred from –S to –S) compared to the recovery of the roots observed in response to supply of sulfate (Figure 1b; plants transferred from –S to +S). The PR growth was slightly enhanced when the seedlings were transferred from the –S preculture to the +S medium (Figure 1a,b). In contrast, the seedlings from the +S preculture medium transferred either to the +S or –S medium showed no significant changes in the root morphological phenotypes (Figure 1c,d).

Based on these observations, we hypothesized the CLE-CLV1 signaling pathway, which has been shown to regulate LR development under N-starved conditions [19], would be involved in pathways modulating root growth under S-limited conditions. To test this hypothesis, we focused on studying the effect of *clv1* mutations on root growth under conditions where the differences in root morphology were most significant. As mentioned, the most significant morphological changes were observed for

the –S-precultured seedlings transferred to either the –S or +S medium (Figure 1a,b). The effect of S supply on changes in root morphology was recorded at Day 7 before the transfer and during the 3 consecutive days in the post-transfer growth period until Day 10 (Figures 2–4). The results indicated that changes in S conditions had little influence on primary root growth (Figure 2). The *clv1-101* mutant had slightly shorter PR values compared to its background wild-type accession (Col-0), regardless of changes in S conditions; however, the same phenotype was not observed in the *clv1-4* mutant in comparison with its background wild-type accession Landsberg *erecta* (L*er*) (Figure 2).

**Figure 1.** Effect of sulfur (S) supply on root phenotypes of *Arabidopsis thaliana* (*A. thaliana*) seedlings. The wild-type Columbia-0 (Col-0) seedlings were grown vertically on –S (15 μM sulfate) or +S (1500 μM sulfate) medium for 7 days and transferred to –S or +S medium in a reciprocal manner to be grown subsequently for 3 days. The scanned images of seedlings on Day 10 are shown. The images show the phenotypes of representative seedlings transferred from (**a**) –S to –S, (**b**) –S to +S, (**c**) +S to +S, and (**d**) +S to –S, respectively.

**Figure 2.** Effect of S supply on primary root growth. Wild-type (Col-0 and L*er*) and *clv1* mutant lines (*clv1-101* and *clv1-4*) were grown vertically on –S (15 μM sulfate) medium for 7 days and then transferred to (**a**) –S medium or (**b**) +S medium to be grown subsequently for 3 days. Primary root length (PRL) was measured before (Day 7) and after the transfer (Day 7+1, 7+2, and 7+3) as indicated on each graph. Values show means (± SE) of 24 individual plants per treatment. White and dark grey bars labeled –S (15 μM sulfate) and +S (1500 μM sulfate) below the graph represent the S conditions before the transfer and during the treatment.

As for the phenotypes associated with LR development, the LR density (LRD; total lateral root number divided by primary root length (cm<sup>−</sup>1)) was maintained at low levels in the wild-type Col-0 seedlings during the period of S limitation which was extended for 3 days after the transfer, while it was restored upon sulfate supply and became higher during the time course (Figure 3). The LRD showed a trend of linear increase over the time course, for which the slope value can be calculated based on linear regression. The slope value of the linear regression line is expressed as the number of LR developed in one cm unit length of PR per day, indicating the daily increase rate of LRD. Thus, it provides a quantitative measure for the assessment of incremental changes in LR development demonstrated over time as a part of the RSA phenotype (Figure 1). The slope values were 0.36 on +S and 0.20 on –S medium, respectively (Figure 3a,b), suggesting that LR emergence which gave rise to visible LR happened 1.8-fold more frequently in Col-0 roots transferred to sulfate-supplied conditions compared to those exposed to prolonged S deficiency. These estimations were generally consistent with the visible phenotypes (Figure 1). Similar changes were observed in the wild-type L*er* seedling; however, the differences between the slope values (0.41 on +S and 0.35 on –S medium, respectively) were not as significant as those estimated for Col-0 (Figure 3a,b).

**Figure 3.** Effect of S supply on lateral root density (LRD). Wild-type (Col-0 and L*er*) and *clv1* mutant lines (*clv1-101* and *clv1-4*) were grown vertically on –S (15 μM sulfate) medium for 7 days and then transferred to (**a**) –S medium or (**b**) +S medium to be grown subsequently for 3 days. LRD was calculated at each time point based on the number of emerged lateral roots (LRs) and the length of the primary root (PR) of one seedling. Values represent means (± SE) of 24 individual plants per treatment. The equations for the linear repression and the R-squared values are indicated on each graph.

Consistent with our previous findings [19], LRD was constantly higher in the *clv1* mutants than in the wild-type (Figure 3). LRD increased significantly in the *clv1* mutants during the period of prolonged S deficiency (Figure 3a; transfer from –S to –S) compared to those transferred to the sulfate-supplied medium (Figure 3b; transfer from –S to +S). In contrast, the wild-type seedlings showed a lower increase rate of LRD on the –S medium compared to those transferred to the +S medium. Based on the slope values of the linear regression, the daily increase rate of LRD was estimated to be 3-fold greater in the *clv1-101* mutant than in Col-0 under S-deficient conditions (Figure 3a). In contrast, upon sulfate supply, the slope value representing the increase rate of LRD was diminished in *clv1-101* though increased in Col-0, apparently converging to the same level (Figure 3b). Similar trends were found when the *clv1-4* mutant was compared with L*er*, while the differences were not so obvious as those shown with the *clv1-101* mutant and Col-0. These results indicated that CLV1 is a signaling component which is necessary for regulation of LR development under S-deficient conditions.

Lateral root branching density (BD) is an alternative measure of LR density, which is calculated by the number of emerged LRs divided by the length of the branching zone (the distance from shoot base to the youngest emerged LR) [29]. BD was calculated 2 and 3 days after the transfer of seedlings to –S or +S medium from 7 days of preculture on the –S medium (Supplemental Figure S1). Consistent with the results shown for LRD (Figure 3), the wild-type plants exposed to prolonged S deficiency showed low BD, while this branching phenotype was recovered upon sulfate supply (Supplemental Figure S1). BD was significantly higher in the *clv1* mutants compared to the wild-type under prolonged S deficiency, while it was lowered after S supplementation. These changes in BD implicate that LR primordia located between the emerged LRs are partially arrested or delayed in progression in the wild-type under the prolonged S deficiency, while their growth can be recovered by the S supplementation. These results also indicate that CLV1 is involved in the regulatory pathway that inhibits the growth of LR primordia, as shown by an increase in BD in the *clv1* mutants relative to the wild-type (Supplemental Figure S1), which confirms our previous findings providing evidence for its essential role in regulating developmental stage progression of LR primordia [19]. Our present findings suggest that long-term S deficiency signals to the CLV1-directed pathway to modulate LR development as demonstrated by changes in LRD and BD.

The total LR length density (TLRLD; total lateral root length divided by primary root length (cm cm<sup>−</sup>1)) was also calculated based on the measurement of the length of the entire LR developed in the root system (Figure 4). The *clv1* mutants showed a significant increase in TLRLD during the period of prolonged S limitation, where an enhanced response to –S was identified in comparison with +S. In contrast, the wild-type showed a more significant increase in TLRLD after sulfate supplementation. These opposing effects of S limitation and sulfate supply on TLRLD in the *clv1* mutants and the wild-type were consistent with those identified for LRD and BD (Figure 3 and Supplemental Figure S1). These results further suggested the essential role of CLV1 in regulation of LR development under S deficiency. As shown in Figure 1, the effect of +S to +S and +S to –S transfer on root morphology was examined simultaneously in this study. In contrast to the results obtained from the –S to –S and the –S to +S transfer experiments (Figures 3 and 4), the LRD and TLRLD of the +S-precultured seedlings changed only slightly in response to S supplementation and S removal (Supplemental Figures S2 and S3). The LRD and TLRLD were higher in the *clv1* mutants than the wild-type as expected, but there was no substantial effect of S in contrast to the phenotypes identified in roots transferred from the –S preculture to –S and +S conditions (Figures 3 and 4).

**Figure 4.** Effect of S supply on total lateral root length density (TLRLD). Wild-type (Col-0 and L*er*) and *clv1* mutant lines (*clv1-101* and *clv1-4*) were grown vertically on –S (15 μM sulfate) medium for 7 days and then transferred to (**a**) –S medium or (**b**) +S medium to be grown subsequently for 3 days. TLRLD was calculated at each time point based on the lengths of the entire LR in one seedling and the length of the PR. Values show means (± SE) of 24 individual plants per treatment.

### *2.2. Regulation of CLE2 and CLE3 Gene Expression under S Deficiency*

To investigate the responses of *CLE* gene expression to the alteration of S nutritional status, the wild-type and *clv1* mutant lines were grown with different S supplies as indicated above for the root phenotype analysis. The root RNAs were extracted at Day 10 (i.e., Day 7+3) for the gene expression analysis. Among the *CLE* genes, *CLE2* and *CLE3* were selected for gene expression analysis, because *CLE2* was previously reported as a S-responsive *CLE* gene [24], and *CLE3* was shown to be significantly up-regulated by N deficiency and inhibits LR development through acting on the CLV1 signaling pathway [19]. *SULTR1;1* encodes a high-affinity sulfate transporter which is highly regulated in response to sulfur deficiency (–S) in the epidermis and cortex of *A. thaliana* roots [30,31]. Therefore, *SULTR1;1* was used as an indicator for tracking the changes in S status. The results indicated prolonged S deficiency (transfer from –S to –S) caused repression, although the sulfate replenishment (transfer from –S to +S) allowed induction of *CLE2* and *CLE3* gene expression in both wild-type and *clv1* mutant lines (*P* < 0.05) (Figure 5a,b). *CLE3* showed greater responses (2–3 fold) to S than *CLE2* (1.2–1.5 fold).

To investigate the effect of S on CLV1-mediated feedback regulation, the *clv1*/wild-type ratios of the *CLE2* and *CLE3* mRNA levels were calculated and compared between the prolonged –S (transfer from –S to –S) and the sulfate supplied (transfer from –S to +S) conditions. The results indicated both the *clv1-101*/Col-0 and *clv1-4*/L*er* ratios of the *CLE3* mRNA levels under the prolonged –S conditions were higher compared to those estimated for roots transferred to +S medium, suggesting that prolonged S deficiency activates a pathway downstream of CLV1 to feedback regulate *CLE3*. In contrast to *CLE2* and *CLE3*, the *CLV1* expression levels did not change significantly by perturbation of S supply (Figure 5c). *SULTR1;1* was upregulated in roots exposed to prolonged S deficiency while repressed upon sulfate replenishment, showing typical patterns of its S-responsive gene expression, which verified that the S conditions tested in this study were appropriate (Figure 5d).

**Figure 5.** Regulation of *CLE2* and *CLE3* transcript levels by S deprivation and sulfate supply. Wild-type (Col-0 and L*er*) and *clv1* mutant lines (*clv1-101* and *clv1-4*) were grown vertically on –S (15 μM sulfate) medium for 7 days and then transferred to –S medium or +S medium to be grown subsequently for 3 days. The mRNA levels of (**a**) *CLE2,* (**b**) *CLE3,* (**c**) *CLV1* and (**d**) *SULTR1;1* in roots at Day 10 (i.e., Day 7+3) were determined by real-time PCR. Roots of wild-type plants grown on the +S medium for 10 days were used as reference samples for relative quantifications. Actin 2 and Ef1α were used as internal controls. Mean values (±SE) of 4 biological replicates with 8 plants for each replicate were calculated using two internal controls. Asterisks indicate statistically significant differences (*P* < 0.05) between gene expression on –S and +S treatment. The S conditions and the order of transfers are shown by white and dark grey bars and arrows below the graph.

### **3. Discussion**

The results shown in this study indicate that the CLE-CLV1 signaling pathway is involved in regulation of LR development under prolonged S deficiency. The CLE-CLV1 signaling module has direct impact on LR development and physiological responses associated with changes in RSA. The proposed model describes three steps of the S-dependent signals and their coordinated actions controlling the CLE-CLV1-dependent pathway, extending our knowledge on S nutrient signaling mechanisms involved in regulation of RSA (Figure 6).

The inhibition of LR development occurs in the wild-type plants exposed to prolonged S deficiency, while this inhibition is abolished in the *clv1* mutants (Figures 3 and 4). In contrast, transferring the seedlings to the S-sufficient medium leads to a recovery of LR development in the wild-type, but rather a diminished response in the *clv1* mutants. The increase rate of LRD calculated based on linear regression provides additional information allowing for a quantitative interpretation of the LR phenotypes, as it is shown to be altered in response to S conditions and different among the genotypes being tested in this study. As described in Figure 3, the daily increase rate of LRD was greater in the *clv1* mutant than in the wild-type during the period of prolonged S deficiency, but estimated to be similar between the *clv1* mutant and the wild-type after the sulfate supply. These findings suggest that CLV1 is a signaling component acting on pathways negatively controlling LR development. Regulatory components expressed downstream of CLV1 may be activated under S-deficient conditions (Figure 6). The number of the emerged and visible LR was counted in experiments performed in this study. Thus, we assume that an increase in LRD after the transfer of precultured seedlings to the new medium

corresponds to the number of newly emerged LR, and it is associated with developmental stage transition of LR primordia—as has been shown previously [19]. The inhibition of LR development under S deficiency and its CLV1 dependency may reflect these developmental mechanisms previously shown with relevance to the N starvation responses [19].

**Figure 6.** Schematic model of signaling pathway controlling lateral root development under S-deficient conditions. S availability affects the activity of the signaling module composed of CLAVATA3 (CLV3)/EMBRYO SURROUNDING REGION (CLE) peptides and CLAVATA1 (CLV1) leucine-rich repeat receptor kinase in roots. The model proposes three potential pathways involved in regulation of LR development under S-deficient conditions. Regulatory components expressed downstream of CLV1 become active under S deficiency and inhibit LR development (**a**). *CLE2* and *CLE3* can be repressed under S deficiency directly or partially through the CLV1-mediated feedback mechanism (at least for CLE3) to reduce the input of CLE signals (**b**,**c**).

*CLE2* and *CLE3* transcript levels decrease under S deficiency in both the wild-type and the *clv1* mutant lines (Figure 5). These transcript profiles suggest that *CLE2* and *CLE3* are repressed under S-deficient conditions in a CLV1-independent manner. In addition, changes in the *clv1*/wild-type ratios of the *CLE3* transcript levels indicate that the CLV1-dependent feedback control mechanism may be strengthened under S-deficient conditions (Figure 5). However, the *CLE2* and *CLE3* transcript levels being reduced in the wild-type plants under S-deficient conditions had no positive impact on promoting LR development. As shown in our model, components expressed downstream of CLV1 are suggested to be more crucial and directly involved in regulation of LR development (Figure 6). The *CLE2* and *CLE3* transcript repression occurring under S-deficient conditions and partially in conjunction with the CLV1-mediated feedback pathway may be considered mechanisms that counteractively reduce the amplitude of the input CLE signals (Figure 6). *CLE2* and *CLE3* are expressed in the pericycle, while CLV1 is expressed in the phloem companion cells [19]. *CLE2* expression is also found in the vascular tissue at the base of developing LR primordia [19,32]. It is suggested that these CLE peptides are SSPs secreted from the pericycle and diffused toward the phloem companion cells to bind to the receptor CLV1, and the CLV1-downstream components expressed in the phloem possibly carry the information to regulate development of LR primordia in long distance via trafficking through the phloem connection [19,33]. The CLE3-CLV1 ligand–receptor relationship and the potential long-distance effect of this signaling module are supported by evidence showing strong inhibition of LR development in transgenic lines with *CLE3* gene overexpression driven by its own promoter in the wild type background and an apparent insensitive response to the same transgene overexpression observed in the *clv1* mutant [19]. Our previous findings provide further implication that additional CLV1-downstream signals may be present and in turn sent back to the pericycle to feedback control *CLE* gene expression [19,33]. According to these models, both CLE peptides and CLV1-downstream components are likely transported between the two different cell types, i.e., pericycle and phloem, and through the vascular system. The actual signals involved in these putative long distance pathways yet remain to be identified.

### **4. Materials and Methods**

### *4.1. Plant Growth Conditions*

Two *Arabidopsis thaliana* accessions, Columbia-0 (Col-0) and Landsberg *erecta* (L*er*), and their corresponding *clv1* mutants, *clv1-101* and *clv1-4* were used in this study. Plants were grown vertically on a nutrient medium containing 1% agar and 1% sucrose as described previously in Reference [19] in growth chambers (CU-36L4; Percival Scientific, Perry, IA, USA) conditioned at 22 ◦C under 16h-light/8h-dark long-day cycles with 75 μmol m−<sup>2</sup> s−<sup>1</sup> light intensity. Agar was washed 6 times with 1 liter of deionized water and remaining water was poured off after each agar settlement to remove sulfate. S-replete (+S) medium contained 1500 μM MgSO4 as the sulfur source. S-deficient (–S) medium contained 15 μM MgSO4, and Mg concentration was adjusted to 1500 μM by adding MgCl2. After 7 d preculture on +S or –S medium, the seedlings were transferred to plates with the same or different concentration of sulfate and grown for 3 d.

### *4.2. Root Analysis*

Roots on agar plates were scanned at Day 7 before transfer and the following 3 days after transfer by using a scanner (Epson Perfection V700 Photo; Seiko Epson, Suwa, Japan) at 300 dpi. The images of roots were traced and measured using ImageJ. LR number, LR length, and PR length were recorded for calculation of LR density (LRD, cm<sup>−</sup>1) and total LR length density (TLRLD, cm cm−1). Total LR length (TLRL, cm) is the sum of the entire lateral root length per plant.

### *4.3. Quantitative Real-Time PCR*

Roots were homogenized by using TissueLyser II (Qiagen, Hilden, Germany). Total RNAs were extracted by using E.Z.N.A. ® Plant RNA Kit (Omega Bio-Tek, Norcross, GA, USA). Turbo DNA-free kit (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) was used for genomic DNA removal of RNA samples. First-strand complementary DNA (cDNA) was prepared from 500 ng of root total RNA by using SuperScript III First-Strand Synthesis System (Invitrogen, Thermo Fisher Scientific). Quantitative real-time PCR was performed by using SYBR Select Master Mix (Applied Biosystems, Thermo Fisher Scientific) on a QuantStudio 7 Flex Real-Time PCR System at the Research Technology Support Facility (RTSF) of Michigan State University. The primers were previously published [19,31,34].

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2223-7747/8/4/103/s1, Figure S1: Effect of S supply on lateral root branching density, Figure S2: Effect of S removal on lateral root density (LRD), Figure S3: Effect of S removal on total lateral root length density (TLRLD).

**Author Contributions:** W.D., and H.T. designed the research. W.D. performed the plant vertical culture, qRT-PCR, and data analysis. Y.W. performed the root measurements using ImageJ. W.D. wrote the paper. H.T. reviewed and edited the paper.

**Funding:** The authors acknowledge funding support from the National Science Foundation (MCB-1244300, IOS-1444549) and the USDA National Institute of Food and Agriculture (Hatch Project 1002395 and 1018575).

**Acknowledgments:** The authors thank Hiroo Fukuda (University of Tokyo) for kindly providing the *clv1* mutant lines for this research.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Sulfate-Induced Stomata Closure Requires the Canonical ABA Signal Transduction Machinery**

### **Hala Rajab 1,2, Muhammad Sayyar Khan 2, Mario Malagoli 3, Rüdiger Hell <sup>1</sup> and Markus Wirtz 1,\***


Received: 23 November 2018; Accepted: 11 January 2019; Published: 16 January 2019

**Abstract:** Phytohormone abscisic acid (ABA) is the canonical trigger for stomatal closure upon abiotic stresses like drought. Soil-drying is known to facilitate root-to-shoot transport of sulfate. Remarkably, sulfate and sulfide—a downstream product of sulfate assimilation—have been independently shown to promote stomatal closure. For induction of stomatal closure, sulfate must be incorporated into cysteine, which triggers ABA biosynthesis by transcriptional activation of NCED3. Here, we apply reverse genetics to unravel if the canonical ABA signal transduction machinery is required for sulfate-induced stomata closure, and if cysteine biosynthesis is also mandatory for the induction of stomatal closure by the gasotransmitter sulfide. We provide genetic evidence for the importance of reactive oxygen species (ROS) production by the plasma membrane-localized NADPH oxidases, RBOHD, and RBOHF, during the sulfate-induced stomatal closure. In agreement with the established role of ROS as the second messenger of ABA-signaling, the SnRK2-type kinase OST1 and the protein phosphatase ABI1 are essential for sulfate-induced stomata closure. Finally, we show that sulfide fails to close stomata in a cysteine-biosynthesis depleted mutant. Our data support the hypothesis that the two mobile signals, sulfate and sulfide, induce stomatal closure by stimulating cysteine synthesis to trigger ABA production.

**Keywords:** sulfate; abscisic acid; stomatal closure; phytohormone synthesis; NADPH oxidase; Protein phosphatases 2C; Sucrose non-fermenting Related Kinase 2 (SnRK2); reactive oxygen species (ROS)

### **1. Introduction**

Plants have to respond to diverse environmental challenges to optimize growth and ensure survival during stress. Regulation of the stomatal aperture is a critically controlled stress response of plants. Stomata are the gates of the plants for interaction with their environment, and various input signals such as pathogen attack, CO2-concentration, light, heat, humidity, and soil water supply, need to be integrated for the optimal opening of these pores. Phytohormone abscisic acid (ABA) is a potent regulator of stomatal aperture and has been shown to transduce many abiotic and biotic input signals for stomatal closure.

The ABA signal transduction cascade is one of the best-characterized input transmission pathways of plants at the molecular level. In guard cells, ABA physically interacts with the PYRABACTIN RESISTANCE1 (PYR1) and PYR1-LIKE (PYL)-proteins or regulatory components of the ABA receptor (RCAR). Binding of ABA to PYR/PYL receptor enhances the affinity of PYR/PYL receptors for ABI1, a PROTEIN PHOSPHATASE of type 2C (PP2C). PP2Cs are inhibited after binding to the ABA-PYR/PYL receptor complex. Inactivation of PP2Cs by ABA causes activation of subclass III Sucrose non-fermenting Related Kinase 2 (SnRK2s) [1], of which SnRK2.6 (OPEN STOMATA 1, OST1, Q940H6) is most relevant for stomatal closure. OST1 is a very potent actor since it phosphorylates several targets whose activities contribute to stomatal closure. One of these targets is the SLOW ANION CHANNEL 1 (SLAC1, [2]). SLAC1-induced current changes result in activation of outward K+ channels. The K+ efflux decreases the osmotic potential in the guard cells, followed by water export. The resulting decreased turgor of the guard cell is the physical cause for stomatal is is closure [3]. Furthermore, OST1 can phosphorylate the plasma membrane-resident β-nicotinamide adenine dinucleotide 2- -phosphate (NADPH) oxidase RESPIRATORY BURST OXIDASE HOMOLOG F (RBOHF, O48538) at Ser<sup>13</sup> and Ser174, which is crucial for the regulation of RBOHF activity [4]. NADPH oxidases produce apoplastic reactive oxygen species (ROS) that are essential for ABA-induced stomatal closure. Internal and apoplastic ROS affect multiple steps during ABA signaling and act as a second messenger that plays a predominant role as an ABA signal amplifier [5].

Several studies connect the drought stress response to the assimilation of sulfur [6,7]. Drought stress regulates the sulfur assimilation pathway in an organ-specific manner and causes differential accumulation of sulfur-metabolism related compounds of the primary sulfur metabolism (e.g., glutathione) and the secondary sulfur metabolism (e.g., 3- -phosphoadenosine 5- -phosphate, PAP) [7,8]. The ROS scavenger glutathione acts in the cytosol, the plastids, and the mitochondria as a redox buffer during stress-induced accumulation of ROS. Its drought-stress induced accumulation has been interpreted as a protection mechanism to avoid over-oxidation of these compartments upon stress-induced ROS formation [9–12].

In contrast, PAP acts as a redox stress-induced retrograde signal of the chloroplast in drought and high light signaling by affecting the expression of nuclear-encoded stress-related genes [8,13]. Recently, PAP has also been shown to act as a second messenger of ABA signaling during stomatal closure that bypasses the canonical ABA signaling components ABI1 and OST1 [8]. PAP is a byproduct of sulfation reactions catalyzed by cytosolic sulfotransferases that transfer the activated sulfate of 3- -phosphoadenosine 5- -phosphosulfate (PAPS) to various compounds [14]. The cytosolic PAP is counter exchanged with the predominantly plastid-generated PAPS and degraded in the plastids by the highly redox-regulated 3- (2- ),5- -bisphosphate nucleotidase (EC 3.1.3.7, SAL1) [13,15].

Remarkably, sulfate is an early xylem-borne chemical signal in maize under soil drying conditions and precedes the root-to-shoot transport of ABA and the pH increase of the xylem sap [16]. ABA transport and an increase of xylem sap pH have long been assumed to serve as a root-to-shoot signal during soil drying. In Poplar, drought-stress also increases the concentration of sulfate in the xylem by lowered sulfate xylem unloading via PtaSULFATE TRANSPORTER 3;3a (PtaSULTR3;3a) and PtaSULTR1;1, and by enhanced sulfate loading from parenchyma cells into the xylem via ALUMINIUM ACTIVATED MALATE TRANSPORTER3b (PtaALMT3b). Furthermore, sulfate has been shown to decrease relative transpiration and stomatal conductance after petiole feeding of sulfate to detached Poplar leaves [17]. The studies established soil-drying induced xylem transport of sulfate as a candidate for the root-to-shoot signal of the water status but did not uncover the signal transduction pathway for sulfate-induced stomatal closure at the molecular level (see below).

After xylem transport of sulfate to the leaves, the sulfate can be stored in the vacuole of leaf cells or transported into the plastids where it can be activated to APS by ATP sulfurylase [18]. APS is either substrate for production of PAPS by APS kinase or reduced to sulfide by subsequent action of APS REDUCTASE (APR) and SULFiTE REDUCTASE (SiR). The competition between APS reductase and APS kinase for their common substrate APS controls sulfur partitioning between the primary and secondary sulfur metabolism [19]. APR and SiR are exclusively localized in plastids and control the flux through the assimilatory sulfate reduction pathway which generates sulfide [20,21]. Three *O*-ACETYLSERINE-THIOL-LYASE (OAS-TL) isoforms catalyze in the plastids, the mitochondria, and the cytosol the incorporation of sulfide into cysteine, which is the precursor for all reduced sulfur-containing compounds in plants, e.g., glutathione [20–24]. The carbon backbone for incorporation of sulfide into cysteine is *O*-acetylserine (OAS), and is separately produced in all

subcellular compartments with its own cysteine synthesis by five SERINE ACETYLTRANSFERASEs (SERATs, [25]). The subcellular localization of OAS-TL and SERAT implies that significant amounts of the membrane-permeable sulfide move from the plastids to the cytosol and the mitochondria for incorporation into cysteine [26,27].

Sulfide is highly toxic and efficiently incorporated into cysteine in mitochondria, which significantly contributes to the detoxification of elevated sulfide levels [28]. On the other hand, sulfide acts in humans and plants as a volatile gasotransmitter that controls various physiological responses [29,30]. In plants, sulfide represses autophagy and induces stomatal closure [31,32]. However, the mode of sulfide-induced stomatal closure is still controversially discussed. Sulfide has been suggested to (1) affect ABA receptor expression directly [33], (2) act upstream of nitric oxide (NO) to modulate ABA-dependent stomatal closure [34], (3) induce in a NO-dependent manner accumulation of 8-mercapto-cGMP for stomatal closure [35], or (4) activate SLAC1 in an OST1- dependent manner [31].

We recently showed that sulfate must be incorporated into cysteine to trigger stomata dynamics. Consequently, sulfate-induced stomata closure was impaired in mutants deficient in the synthesis of cysteine or ABA [36]. Remarkably, cysteine synthesis depleted mutants are sensitive to drought and high light stress [36–38]. Both stresses also cause PAP accumulation. Since sulfide is a downstream product of assimilatory sulfate reduction pathway and PAP formation is a result of sulfation reactions, it is important to dissect how PAP, sulfide, and sulfate control stomatal aperture.

Here, we apply reverse genetics to understand the contribution of the canonical ABA signaling machinery to sulfate-induced stomata closure and dissect the potential relevance of the sulfation byproduct PAP in this process. We found that the protein phosphatase ABI1 and the down-stream kinase OST1 are essential for sulfate-induced ROS formation in stomata and stomatal closure. Since PAP acts independently of OST1, we concluded that potential accumulation of PAP upon external sulfate administration does not significantly contribute to sulfate-induced stomatal closure. In concordance with the function of ROS as an amplifier of ABA signaling, the loss-of-function mutants for the NADPH oxidases RBOHD and RBOHF are also impaired in sulfate-induced stomatal closure. We furthermore demonstrate that sulfide-induced stomatal closure requires the presence of the major SERAT isoforms located in the cytosol, the plastids, and the mitochondria, strongly suggesting that sulfide needs to be integrated into cysteine to promote stomatal closure. We suggest that sulfate and sulfide are incorporated into cysteine to trigger ABA formation, which in turn requires canonical ABA signaling components to mediate sulfate/sulfide/cysteine-induced stomatal closure.

### **2. Results**

In our previous study, we demonstrated that sulfate can close stomata and that it needs to be incorporated into cysteine for activation of ABA synthesis and accumulation of ABA in the cytosol of guard cells. However, we did not show how sulfate-induced ABA is perceived to trigger stomatal closure.

### *2.1. Sulfate-induced Stomatal Closure Requires Functional ABA Signaling*

In order to provide direct evidence for the biological relevance of the ABI1 during sulfate-induced stomata closure, we challenged epidermal peels of 5-week-old soil grown wild-type and *abi1-1* plants with 15 mM sulfate for three hours. In the *abi1-1* mutant, the protein phosphatase ABI1 is constitutively active, which results in a permanent open stomata phenotype ([39,40], Figure 1a). Application of 15 mM sulfate efficiently closed the stomata of wild-type plants (Figure 1a), which supports the previously reported impact of sulfate on stomatal closure [17,36]. In contrast, stomata did not close in the *abi1-1* mutant after application of sulfate, demonstrating that sulfate-induced ABA accumulation in guard cells is not triggering stomatal closure when the PYR/PYL-ABA-ABI1 ternary complex is non-functional (Figure 1a,b).

**Figure 1.** Functional ABI1 and OST1 are essential for sulfate-induced stomatal closure in Arabidopsis. (**a**) Representative guard cell in epidermal peels of 5-weeks–old soil-grown wild-type, *abi1-1*, and *ost1-2* plants that were floated on water at pH 5.5 or water supplemented with sulfate (15 mM MgSO4) for three hours. (**b**) Quantification of the stomatal aperture of guard cells in epidermal peels from plants indicated in (**a**) that were treated with water (control, black) supplemented with sulfate (15 mM, yellow). Letters indicate statistically significant differences between groups determined with the one Way ANOVA test (*p* < 0.05, n = 50 stomata, derived from epidermal peels of five individual plants). Values represent means ± standard deviation (SD). Scale bar, 10 μm.

In a separate experiment, we independently confirmed the absence of sulfate responsiveness for *abi1-1* and applied ABA as a control for stomatal closure. Treatment of sulfate (15 mM) and ABA (50 μM) resulted in significant stomatal closure in the wild-type. The degree of stomatal closure was indistinguishable after sulfate and ABA application. ABA or sulfate application to epidermal peels of the *abi1-1* mutant did not affect the stomatal aperture (Figure 2).

**Figure 2.** Functional ABI1 is required for stomatal closure upon ABA or sulfate treatment. (**a**) Representative guard cell in epidermal peels of 5-week–old soil-grown wild-type and *abi1-1* plants that were floated on water at pH 5.5 or water supplemented with sulfate (15 mM MgSO4) or ABA (50 μM) for three hours. (**b**,**c**) Quantification of the stomatal aperture in epidermal peels of the wild-type (**b**) and the *abi1-1* mutant (**c**) that were treated with water (control, black) supplemented with sulfate (15 mM, yellow) or ABA (50 μM, white). Letters indicate statistically significant differences between groups determined with the one Way ANOVA test (*p* < 0.05, n = 50 stomata, derived from epidermal peels of five individual plants). Values represent means ± standard deviation (SD). Scale bar, 10 μm.

Binding of ABA to PYR/PYL receptors causes efficient inactivation of the PP2C named ABA INSENSITIVE1 (ABI1). The inactivation of ABI1 releases the downstream kinase OST1 from inhibition, which in turn phosphorylates multiple targets to promote ABA-induced stomatal closure. In the *ost1-2* mutant, a T-DNA insertion in the OST1 gene destabilizes the OST1 mRNA, resulting in a total loss-of-OST1 function. Like the *abi1-1* mutant, *ost1-2* displays constitutively open stomata and is sensitive to soil drying and low humidity ([41], Figure 1a). The absence of OST1 kinase activity inhibits the impact of sulfate on stomatal aperture (Figure 1a,b), which is in concordance with the previously demonstrated function of sulfate for the promotion of ABA accumulation in guard cells [36]. These results suggest that the PYR/PYL receptors sense sulfate-induced accumulation of ABA in guard cells and support the importance of the ABI1-OST1 phosphorylation relay for sulfate-induced stomatal closure.

### *2.2. ABI1 and OST1 are Essential for the Sulfate-induced Formation of ROS in Guard Cells*

A known target of the ABI1-OST1 phosphorylation relay is the membrane resident NADPH oxidase RBOHF, which is essential for ABA-induced ROS production [4]. We, therefore, tested the formation of ROS in guard cells after sulfate-application to epidermal peels from the wild-type, the *abi1-1*, and the *ost1-2* mutant. Application of sulfate (15 mM) resulted in significant production of ROS in the wild type. The degree of ROS formation in response to sulfate was comparable to the formation of ROS after application of 50 μM ABA (Figure 3a,b). The guard cells in epidermal peels of *abi1-1* and *ost1-2* displayed wild-type like levels of ROS under control conditions. Both mutants of the ABI1-OST1 phosphorylation relay failed to produce ROS in response to the application of sulfate, which is in concordance with the failure to close stomata in response to sulfate. Also, ABA application did not induce ROS formation in both mutants under their applied conditions, which is in agreement with results of previous studies [5].

### *2.3. Sulfate Stimulus Activates NADPH Oxidases for Production of ROS in Guard Cells*

The sulfate-induced accumulation of ROS in wild-type guard cells has also been observed in our previous study on sulfate-induced stomatal closure [36]. In this study, we demonstrated that inhibition of oxidases, like RBOH isoforms A-F, with the selective inhibitor diphenyleneiodonium prevents the sulfate-induced formation of ROS in guard cells. Co-application of sulfate and diphenyleneiodonium also impaired sulfate-induced stomatal closure, strongly suggesting that formation of ROS is vital for transduction of the sulfate stimulus. These findings support the hypothesis that sulfate induces synthesis of ABA in guard cells, which causes the formation of the ABA-PYR/PYL-ABI1 ternary complex resulting in activation of OST1. The activated OST1 potentially phosphorylates NADPH oxidases to produce the second messenger ROS. The elevated ROS levels will act as a signal amplifier to activate SLAC1, leading to stomatal closure upon sulfate stimulus [5].

In order to link the activation of OST1 and the formation of ROS upon sulfate stimulus with the membrane resident NADPH oxidases, we tested the contribution of two major isoforms of RBOHs expressed in guard cells, RBOH-D and RBOH-F, to sulfate-induced ROS formation [42]. The *rboh-D* and *rboh-F* mutant lack functional isoforms of the NADPH oxidase due to a T-DNA insertion in the respective gene. The absence of either RBOH-D or RBOH-F impaired ROS formation in guard cells upon application of ABA or sulfate for three hours to epidermal peels (Figure 4a,b). Consequently, sulfate did not affect stomatal aperture in both mutants (Figure 4c). Wild-type guard cells produced ROS and closed stomata upon sulfate application.

**Figure 4.** Sulfate-induced stomatal closure and ROS formation require the NADPH oxidases RBOHD, and RBOHF (**a**) Visualization of reactive oxygen species (ROS) with the H2O2-selective dye 2- ,7- -dichlorofluorescein (H2DCF) as described in material and methods. Epidermal peels of 5-weeks–old soil-grown wild-type, *rboh-D*, and *rboh-F* plants were floated on water at pH 5.5 or water supplemented with sulfate (15 mM MgSO4) or ABA (50 μM) for three hours prior staining of ROS. (**b**) Quantification of ROS staining in guard cells floated on water (control, black) supplemented with sulfate (15 mM, yellow) or ABA (50 μM, white). Letters indicate statistically significant differences between groups determined with the one Way ANOVA test (*p* < 0.05, n = 50 stomata, derived from epidermal peels of five individual plants). Values represent means ± standard deviation (SD). (**c**) Representative guard cells embedded in epidermal peels of the wild-type and the *rboh-D*, and *rboh-F* mutants after treatment with water or water supplemented with sulfate (15 mM). Scale bar, 10 μm.

### *2.4. Stomata of the Serat tko Mutant do not Close upon Sulfide Application*

The volatile signal H2S is a downstream product of sulfate assimilation and has been shown independently to induce stomatal closure [34,43,44]. Since sulfate must be incorporated into cysteine to gain competence as an inducer of stomatal closure, we decided to test if sulfide is also incorporated into cysteine for induction of ABA biosynthesis. Sulfide is the direct sulfur-precursor of cysteine and is incorporated by the activity of OAS-TL into cysteine. Cysteine biosynthesis is not limited by OAS-TL activity, but by the formation of the carbon skeleton for cysteine, OAS. The three major SERAT

isoforms SERAT1;1, SERAT2;1 and SERAT2;2 produce the bulk of OAS for the incorporation of sulfide into cysteine in the cytosol, the plastids, and the mitochondria. A SERAT triple knock-out mutant (*serat tko*) lacking these major SERAT isoforms is retarded in growth and displays significantly lowered translation due to decreased production of OAS and cysteine [45]. When we applied 100 μM sulfide (NaHS) dissolved in water at pH 5.5 to epidermal peels of the wild-type, the dissolved sulfide triggered the closure of the wild-type stomata within 90 min. The application of water at pH 5.5 for the same duration did not affect the stomatal aperture (Figure 5). Remarkably, sulfide failed to close the stomata in epidermal peels of the *serat tko* mutant. These results suggest that sulfide is not perceived by a membrane-resident receptor located at the plasma membrane, but is incorporated into cysteine to gain competence as a stomatal closure signal.

**Figure 5.** Sulfide is unable to induce stomatal closure in the OAS biosynthesis-depleted *serat tko* mutant. (**a**) Representative guard cells embedded in epidermal peels of the wild-type and the *serat tko* mutant after treatment with water at pH 5.5 (Control) or water supplemented with sulfide (100 μM NaHS) at pH 5.5 for 90 minutes. (**b**) Quantification of the stomatal aperture in epidermal peels of the wild-type and the *serat tko* mutant that were treated with water (control, black), or water supplemented with sulfide (100 μM NaHS, red). Letters indicate statistically significant differences between groups determined with the one Way ANOVA test (*p* < 0.05, n = 50 stomata, derived from epidermal peels of five individual plants). Values represent means ± standard deviation (SD). Scale bar, 10 μm.

### **3. Discussion**

### *3.1. Sulfate and Sulfide are Incorporated into Cysteine to Trigger Stomatal Closure*

Stomatal closure is a dynamic process that optimizes carbon dioxide uptake with transpiration-mediated water loss. A multifaceted signaling network regulates stomatal movements and integrates diverse environmental and endogenous inputs. Some of these inputs are locally generated, e.g., pathogen-induced stomatal closure, and cause fast responses that use phosphorylation relays to control plasma-membrane resident NADPH oxidases [5]. In other cases, e.g., perceiving of the soil-water status, a distantly originated signal must travel to the guard cell and will be integrated to modulate stomatal aperture. Many of these integration processes impinge on the ABA signal transduction pathway [3,46]. However, our knowledge of the regulation of the tissue-specific ABA biosynthesis sites in response to environmental cues is currently quickly evolving, but still limited [3,46].

Characterization of the plastidic sulfate transporter SULTR3;1 uncovered for the first time a direct link between the actual sulfur supply and ABA biosynthesis in plants [47,48]. Indeed, sulfate has been shown by independent groups to accumulate in the xylem sap of drought-stress maize, Poplar, common hop and Arabidopsis plants and was supposed to act as an early signal that promotes ABA-induced stomatal closure [16,17,49,50]. The recent identification of cysteine as a trigger for ABA biosynthesis in guard cells linked the sulfate-induced stomatal closure with the biosynthesis of this factor for stomatal closure [36]. The presented findings establish ABI1 as a required transducer of

the ABA signal during sulfate-induced stomatal closure (Figure 1), which is in agreement with the stimulation of ABA production by cysteine and sulfate in guard cells and rosette leaves [36].

After perception, the ABA signal is transduced via the ABI1-OST1 phosphorylation relay to multiple downstream effectors. In concordance with this canonical view on ABA perception, OST1 is also vital for sulfate-induced stomatal closure. Both results strongly support the view that sulfate stimulates ABA formation and point against a direct gating of the QUICK ANION CHANNEL 1 (QUAC1 or ALMT12) channel by sulfate, which has been hypothesized by Malcheska and co-workers as the molecular basis of sulfate-induced stomata closure. This hypothesis was based on the stimulating effect of sulfate on QUAC1 expressed in *Xenopus* oocytes and the absence of stomatal closure in the *quac1* mutant [17]. The presented findings are not in contradiction with the insensitivity of the *quac1* mutant towards sulfate as a trigger of stomatal closure [17], but offer a different interpretation of the *quac1* insensitivity towards sulfate. Like SLAC1, QUAC1 is required for ABA-induced stomatal closure and is a substrate of OST1, which can activate SLAC1 and QUAC1 by direct phosphorylation [2,51,52]. Consequently, sulfate fails to close stomata in *quac1* [17] and *slac1-1* [36], although SLAC1 is not gated by sulfate.

Sulfide is a well-established signal for stomatal closure. Like sulfate, sulfide can be incorporated into cysteine. The failure of sulfide to stimulate stomatal closure in the *serat tko* (Figure 5), which is impaired in the incorporation of sulfide into cysteine, suggests that sulfate and sulfide use the same mechanism for induction of stomatal closure. Both signals stimulate the synthesis of cysteine, which in turn promotes ABA formation. In support of this hypothesis, sulfide is known to be immediately incorporated into cysteine after short-term exposure [27], and stimulates SLAC1 by activation of OST1 [31]. Furthermore, sulfate fails to close stomata in the *sir1-1* mutant that is depleted in its capability to reduce sulfate to sulfide [36,53].

### *3.2. The Role of ROS as Second-messenger in Sulfate-induced Stomatal Closure*

ROS are important signal amplifiers of the ABA signal and act downstream of ABA as second messengers during stomatal closure and systemic acquired acclimation [5,54]. Like sulfide [55], sulfate application also triggered ROS production in guard cells of the wild type (Figure 3) in an RBOH dependent manner (Figure 4). The formation of ROS is essential for sulfate-induced stomatal closure [36]. The ABA-triggered ROS production in guard cells is OST1 dependent [41], which can directly phosphorylate plasma membrane-resident NADPH oxidases, like RBOH-F [4]. In concordance with the established scheme for ABA-induced ROS formation, sulfate-induced ROS production was diminished in *abi1-1*, *ost1-2* and loss-of-function mutants for NADPH oxidases (RBOH-D and RBOH-F). Remarkably, ABA-triggered formation of ROS depends on ABI1 but not on ABI2 [56]. Indeed, *abi2-1* mutants accumulated ROS in response to sulfate application [36]. Thus, the here presented data strongly suggest that the sulfate-induced ROS formation is a consequence of sulfate-promoted ABA formation, which in turn stimulates membrane resident NADPH oxidases via the classical ABA signal transduction cascade.

### *3.3. Contribution of Cytosolic Sulfation Reactions Releasing PAP to Close Stomata in Response to Sulfate*

PAP is a side-product of sulfation reactions that use PAPs as sulfate donors [14]. PAP is also a potent inducer of stomatal closure [57]. Consequently, one could have hypothesized that sulfate-triggered PAP accumulation might contribute to sulfate-induced stomatal closure. Remarkably, PAP bypasses the canonical ABA-induced stomatal closure pathway and is independent of ABI1 and OST1 [57]. In contrast, sulfate-induced stomatal closure requires functional ABI1 and OST1 (Figures 1–3) and its downstream targets (RBOH-F, Figure 4). Our findings exclude a significant contribution of PAP during sulfate-induced stomatal closure. Accumulation of PAP upon sulfate application is also highly questionable, since PAP levels are strongly regulated by the PAP degrading enzyme, SAL1 [6,8]. SAL1 is highly regulated by environmental stimuli like drought and high-light stress and tightly controls PAP level, which is the basis for the well-established retrograde signaling function of PAP [8,13].

In conclusion, our results uncover that sulfate-induced formation of the stomatal closure signal, ABA, is transduced by ABI1, which has been previously shown to physically interact with the PYR/PYL receptor in an ABA-dependent manner. The canonical ABI1-OST1 phosphorylation relay is essential for the activation of plasma-membrane resident NADPH oxidases of the RBOH type. These RBOHs produce the ABA signal amplifier ROS, which is mandatory for sulfate-induced stomatal closure. The failure of sulfide to close stomata in the *serat tko* mutant supports the view that sulfate and sulfide must be incorporated into cysteine to gain competences as stomatal closure signals due to the stimulation of ABA production.

### **4. Materials and Methods**

### *4.1. Plant Material and Growth*

Seeds of *Arabidopsis thaliana* Col-0 (ecotype Columbia) and the mutants *abi1-1* (AT4G26080) CS22, *ost1-2* (AT4G33950), *rboh-D* (AT5G47910) CS9555, *rboh-F* (AT1G64060) CS68522, *serat tko* (SALK\_ 050213 x SALK\_ 099019 x Kazusa\_KG752, [45]), were sown on soil (Tonsubstrat from Ökohum, Herbertingen) supplemented with 10% (v/v) vermiculite and 2% quartz sand. Seeds were stratified at 4 ◦C for two days in the dark. The plants were grown under long-day conditions for five weeks before the experiment (16 h light, 100 μmol m−<sup>2</sup> s−<sup>1</sup> at 22 ◦C and eight h dark at 18 ◦C for day and night, respectively). Relative humidity was set at 50%.

### *4.2. Stomatal Aperture Bioassay*

Epidermal peels were prepared from the abaxial side of Arabidopsis leaves as described in [16] and allowed to float on distilled water for 2 hours under constant light. The peels were then transferred to distilled water pH 5.5 supplemented without (control) or with effectors (15mM MgSO4 and 50 μM ABA) for the indicated periods. Images of the stomata were captured after the treatment with a conventional wide-angle microscope (Leica DMIRB). The stomatal aperture was determined with ImageJ (http://fiji.sc/), using a μm ruler for calibration.

### *4.3. H2O2 Quantification in Guard Cells*

ROS were determined in intact stomata of epidermal peels from the abaxial side of the leaf according to [58]. The peels were allowed to float on water without and with effectors for up to three hours. Subsequently, the epidermal peels were stained with 50 μM 2- ,7- -dichlorodihydrofluorescein diacetate (H2DCFA) for 10 min and transferred to a microscope slide. The ROS-specific fluorescence was detected at 525 nm after specific excitation at 488 nm using a confocal microscope (Nikon A1R) according to [59]. Images were captured from peels of five individual plants, and the fluorescent signal of intact stomata (50) was quantified using the open source software ImageJ (http://fiji.sc/).

### *4.4. Statistical Analysis*

The Statistical analysis for the experimental data was done using SigmaPlot 12.5 (Systat Inc., San Jose, CA, USA). The analysis of all data sets was analyzed through One Way Repeated Measures Analysis of Variance (one-Way ANOVA) for all of the multiple pairwise comparisons using the Holm-Sidak method. The Shapiro-Wilk method (p to reject was *p* > 0.05) was used to test the normality distribution of data. In the figures, letters are used to indicate the significance difference (*p* < 0.05).

**Author Contributions:** H.R. performed the stomata closure experiments and ROS quantification. M.S.K., M.M., R.H. and M.W. supervised H.R. M.W. and R.H. designed the study. All authors participated in writing the manuscript.

**Funding:** This research was funded by the German Research Association, grant number (HE1848/14-1, -/15-1,-16/1 and WI3560/1-1, -/2-1. H.R. received funding from the Higher Education Commission (HEC)-Pakistan, and the German Academic Exchange Service (DAAD).

**Acknowledgments:** The authors want to thank Prof. Rainer Hedrich (University of Wuerzburg, Germany) and Dietmar Geiger (University of Wuerzburg, Germany) for the kind gift of *abi1-1*, *ost1-2*, *rboh-D* and *rboh-F* mutants. We are grateful for the Higher Education Commission (HEC)-Pakistan, and the German Academic Exchange Service DAAD for their financial support.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
